lable at ScienceDirect

Quaternary International xxx (2014) 1e11

Contents lists avai

Quaternary International

journal homepage: www.elsevier .com/locate/quaint

Late Middle Palaeolithic surface sites occurring on dated sedimentformations in the Thar Desert

James BlinkhornPACEA, Université Bordeaux 1, Bâtiment B18, Avenues des Facultes, 33405 Talence Cedex, France

a r t i c l e i n f o

Article history:Available online xxx

E-mail address: j.blinkhorn@pacea.u-bordeaux1.fr.

1040-6182/$ e see front matter � 2014 Elsevier Ltd ahttp://dx.doi.org/10.1016/j.quaint.2014.01.027

Please cite this article in press as: Blinkhorn,Quaternary International (2014), http://dx.d

a b s t r a c t

The dominance of erosional sedimentary processes in desert regions results in Palaeolithic records rich insurface scatters. Engaging with this abundant archaeological resource, including incorporating themwithin absolute chronologies or schemes of environmental variability, is critical to understandinghominin habitation in arid landscapes. Extensive archaeological surveys in the Thar Desert of India andPakistan over the past 60 years have identified a large number of Palaeolithic surface sites in a diverserange of geographic contexts. In the past twenty years, a rich Upper Pleistocene palaeoenvironmentalrecord has been developed within the region due to the application of chronometric dating techniques.However, as yet there have been few attempts by archaeologists to survey previously dated sedimentformations in the Thar Desert and capitalise upon the potential to attribute preliminary terminus postquem dates to surface sites. Surface survey has been undertaken targeting sediment deposits that maypreserve evidence for hominin behaviour between Marine Isotope Stage 5 and 3, a period in whichtechnological and demographic overhauls in hominin populations appear to occur in South Asia.Following brief review of the Palaeolithic archaeology and palaeoenvironmental history of the TharDesert, the results of recent archaeological survey of dated sediments at Chamu, Karna and Shergarh Tri-Junction are presented. The results offer tentative evidence for the presence of late Middle Palaeolithicpopulations in the region, which are evaluated and discussed within the context of contemporary oc-cupations across southern Asia.

� 2014 Elsevier Ltd and INQUA. All rights reserved.

1. Introduction

The desert environments of the mid-latitude arid belt yieldwidespread evidence for occupation by Middle Palaeolithic pop-ulations during the Upper Pleistocene and share climatic charac-teristics that led Finlayson (2013:30) to label them ‘crucibles ofhuman evolution’. Recent research across northern Africa(Arzarello et al., 2013; Beyin, 2013; Cancellieri and di Lernia, 2013;Foley et al., 2013; Scerri, 2013a,b), Arabia (Petraglia et al., 2011; Roseet al., 2011; Delagnes et al., 2012; Groucutt and Petraglia, 2012;Delagnes et al., 2013; Usik et al., 2013) and India (Blinkhorn,2013; Blinkhorn et al., 2013) supports this suggestion, indicatingthe shear wealth of Palaeolithic sites in these regions. Emergingtheoretical perspectives based in biogeographic approaches pri-oritise the colonisation of the mid-latitude arid belt as critical tounderstanding the Upper Pleistocene dispersals of modern humansin place of the more simplistic dichotomy evident in ‘out of Africa’approaches (Petraglia et al., 2012; Boivin et al., 2013; Finlayson,

nd INQUA. All rights reserved.

J., Late Middle Palaeolithic suoi.org/10.1016/j.quaint.2014.0

2013; Groucutt and Blinkhorn, 2013). In this context, the TharDesert of south east Pakistan and west India (Fig. 1) plays a crucialrole in the investigation of the capabilities of human populations toadapt to radically new environments, as it is the last of the desertlandscapes that would have been colonised by eastward dispersinggroups through the mid-latitude arid belt before entering themonsoonal Oriental biogeographic realm beyond. Therefore, thenature of lithic industries that occur in the Thar Desert, the timingof their appearance and evolution can offer critical insights into thetechnological repertoire and expertise of the first modern humanpopulations and their adaptability in colonising the significantlydifferent habitats encountered in the Indian subcontinent.

The general scarcity of caves or other such sediment traps inmany areas of the mid-latitude arid belt stands in contrast to otherregions with rich Middle Palaeolithic and Middle Stone Age ar-chives such as the Maghreb, the Levant or South Africa’s Westernand Eastern Cape. Instead, erosional sedimentary processes pre-dominate, whether as deflation of fine sediments through aeoliantransportation or the mobility of a broader range of depositsthrough fluvial activity. The result is a highly visible archaeologicalrecord on exposed sediment surfaces. The integration of evidence

rface sites occurring on dated sediment formations in the Thar Desert,1.027

Fig. 1. Map showing location of sites reported in this study (circles) and sites reported with chronometric dates by previous researchers (triangles).

J. Blinkhorn / Quaternary International xxx (2014) 1e112

from such surface sites with that from excavated and dated as-semblages can be difficult. However, where the sediment forma-tions upon which surface sites occur can be dated it is possible tosuggest robust terminus post quem dates for such assemblages.

Contemporary archaeological debate surrounding the modernhuman colonisation of the Indian subcontinent is focused uponwhether an abrupt or gradual transition from Middle to LatePalaeolithic industries occurred at the onset Marine Isotope Stage(MIS) 3 (Boivin et al., 2013; Mellars et al., 2013; Mishra et al., 2013).An abrupt transitionwould suggest that Late Palaeolithic industrieswere requiredbyhumanpopulations to colonise themosaic habitatsof the Oriental biogeographic zone and out-compete residentarchaic hominins (Mellars et al., 2013; Mishra et al., 2013). Incontrast, a gradual transition may indicate that human populationswere able to successfully adapt to these new environments usingMiddle Palaeolithic technologies, with the localised development ofLate Palaeolithic industries a long term consequence of such on-going adaptations (Boivin et al., 2013). Currently, there is littleclear evidence from the Thar Desert as to the nature of lithic in-dustries that existed in MIS 3, whereas Middle Palaeolithic assem-blages occur at the MIS 4/3 boundary and the earliest reported LatePalaeolithic industries occur in an MIS 2 context (Blinkhorn, 2012).An appraisal of surface assemblages that appear on sediment for-mations that post-dateMIS 3 in the Thar Desert offers a preliminarymeans to assess whether Middle Palaeolithic occupations extendedinto MIS 3, if Late Palaeolithic technologies are readily apparent inMIS 3 and the nature of the relationship between the two industries.Such an assessment will help to illuminate the technological meansby which modern human populations were able to transcendbiogeographic boundaries and colonise the Oriental zone.

This paper reports the results of surface surveys of dated sedi-ment formations in central Rajasthan, India. Firstly, a briefgeographic, palaeoenvironmental and archaeological backgroundof this region are set out. Following this, a review of publishedpalaeoenvironmental and chronometric research from three sitesin the Thar Desert is provided. The methodology details the surveystrategy and attributes of lithic artefacts recorded, which are sub-sequently presented in the results section. The ensuing discussions

Please cite this article in press as: Blinkhorn, J., Late Middle Palaeolithic suQuaternary International (2014), http://dx.doi.org/10.1016/j.quaint.2014.0

and conclusion evaluate the effectiveness of the survey to addresshuman adaptations and dispersals in the Thar Desert and con-textualise them within contemporary debate.

2. Geography and palaeoenvironments

The current extent of the Thar Desert is w3,00,000 km2,occurring predominately in the Indian states of Rajasthan, Gujarat,Haryana and Punjab, and the Pakistani provinces of Upper andLower Sindh. The Thar Desert is characterised by low topographicrelief, with marked changes to this observed on three sides (East,North and West), namely the Aravalli range, the foothills of theHimalaya and the Iranian Plateau. To the south, the desert isbounded by the coastline of the Arabian Sea. Low hills occur inplaces, where the regions Tertiary geology becomes exposed, butextensive sedimentation throughout the Quaternary period,exceeding 300 m depth in some places (Bajpai et al., 2001), hasevened out major breaks of slope.

The Indian Summer Monsoon (ISM) plays a critical role in theregion, delivering the majority of the Thar Desert’s precipitationwithin the summer months, whereas the weaker, winter monsoonoften delivers little or no rainfall to this area. As monsoonal circu-lation involves cross-equatorial exchanges of atmospheric heat andpressure, changes tobothArctic andAntarctic ice volumes, aswell asnorthern and southern hemispheric temperatures affect ISM in-tensity. Although maximum monsoonal intensity appears to coin-cide with glacial minima observed at both poles, steep inter-polartemperature gradients appear to typically produce a monsoonalminimum ca. 20 kyrs before the glacial maximum (An et al., 2011;Liu, 2011). As this temperature gradient decreased toward theglacial maximum, monsoonal intensity increased (An et al., 2011;Liu, 2011). Therefore, global proxies for palaeoenvironmentalchange in the region suggest that while enhanced humidity in thisarea would have occurred during the Last Interglacial, the TharDesert may have experienced depressed monsoonal activity be-tween w40 and 20 ka, which, due to the lack of precipitation, mayhave led to extinction of hominin populations in the region. Criti-cally, regional precipitation gradients based upon terrestrial

rface sites occurring on dated sediment formations in the Thar Desert,1.027

J. Blinkhorn / Quaternary International xxx (2014) 1e11 3

palaeoenvironmental proxies are likely toprovide themost accurateinsight into desert margin shift (Singhvi and Kar, 2004).

Few dated terrestrial proxies from the Thar Desert offer insightsinto patterns of palaeoenvironmental change that extend into theMiddle Pleistocene. Evidence for aeolian activity extending intoMIS 6 is present at 16R Dune (Achyuthan et al., 2007; Singhvi et al.,2010), Chamu (Dhir et al., 2010), and Karna (Jain et al., 2005).However, undated deposits underlying Upper Pleistocene sedi-ments clearly indicate periods of significant humidity, such as inthe lacustrine deposits from Singi Talav and fluvial horizons sur-rounding the Jayal gravel ridge (Misra, 1995). In contrast, a range ofproxy data is available for the Upper Pleistocene.

Overall, the Last Interglacial period (MIS 5) was marked by pe-riods of humidity that are likely to have been more humid thanconditions during the early Holocene climatic optimum (Enzelet al., 1999). In the central Thar, evidence exists for strong fluvialactivity, seen in extensive sedimentation in the Orsang, Sabarmatiand Mahi valleys, as well as the presence of braided gravel bedchannels in the Mid Luni Valley (see below) (Jain et al., 2005).Limited dune formation is reported during MIS 5, suggestingintermittent periods of increased aridity in the region, such as at16R Dune and Chamu (Misra et al., 1982; Achyuthan et al., 2007;Dhir et al., 2010; Singhvi et al., 2010). However, the developmentof pedogenic carbonateswithin these horizons indicate exposure toseasonal humidity, and isotopic studies suggest the local vegetationwas dominated by C4 types (Dhir et al., 2010; Singhvi et al., 2010)that thrive during periods of increased ISM intensity (Achyuthanet al., 2007). However, towards the end of MIS 5 there is evidencefor significant reduction in fluvial activity, both in the central Tharand its southern margins (Juyal et al., 2003, 2006; Jain et al., 2005).

Evidence for palaeoenvironments during MIS 4 are somewhatlimited, and where deposits are dated to this period they appearinverted with deposits from the earliest periods of MIS 3. Examplesof this occur at Katoati (Blinkhorn et al., 2013) and Shergarh Tri-Junction (Andrews et al., 1998). With the onset of MIS 3 bothfluvial and aeolian activity are apparent in the Thar. Intermittentaeolian activity during MIS 3 is predominately known from dateddune sites in central Rajasthan, with aggradational events occurringbetween 58 and 36 ka (Chawla et al.,1992; Andrews et al.,1998; Dhiret al., 2010; Singhvi et al., 2010). Fluvial aggradation is also apparentat the Thar’s southern margin, with heavily pedogenised overbankdeposits interstratified with gravel deposits suggesting greaterstability than in earlier periods (Juyal et al., 2006).

The transition between MIS 3 and MIS 2 is marked by sharpdecline in fluvial strength in and the onset of widespread aeoliandeposition in the region (Jain et al., 2005; Juyal et al., 2006) sug-gesting a significant change to the regional precipitation gradients. Itis notable that Sambhar Lake, the largest playa in the region, appearsto have retained a body of standing water over the Last GlacialMaximum(Sinha et al., 2006). Immediately following the Last GlacialMaximum, aeoliandepositionbecomeswidespread across Rajasthanand Gujarat (Chawla et al., 1992; Srivastava et al., 2001; Juyal et al.,2006; Dhir et al., 2010; Singhvi et al., 2010) and fluvial activity isresumed (Jain et al., 2005; Juyal et al., 2006). This combination ap-pears to have led to the formation of a number of playas in fluvialcourses whose drainage is blocked by sand dunes, such as BapMalarand Talchappar (Deotare et al., 2004; Achyuthan et al., 2007).

3. Archaeological background

Scholarly interests in the Thar Desert during the first half ofthe 19th century were predominantly geological and geograph-ical, after which collections of stone age artefacts began to bemade sporadically from the 1860’s onwards (Allchin et al., 1978).Systematic research began in the region following independence

Please cite this article in press as: Blinkhorn, J., Late Middle Palaeolithic suQuaternary International (2014), http://dx.doi.org/10.1016/j.quaint.2014.0

from British rule, and the subsequent delineation of states withinIndia and it’s separation with Pakistan appears to have hadconsiderable influence upon the distribution of survey work.Following Zeuner (1950), a number of researchers in the regionhave situated both surface and buried archaeological depositswithin their environmental context using relative stratigraphyand chronology, of which Allchin and colleagues’ (1978) syn-thesis offers a superb example of the value of such approacheswhere numerical chronologies are lacking. Recent appraisal ofpublished data on Palaeolithic sites in the Thar Desert identified939 separate reports (Blinkhorn, 2012), indicating the region isindeed rich in Palaeolithic heritage, although the data providedby the majority of reporters is difficult to compare at an inter-regional scale.

The evidence that is available from surface sites in the TharDesert indicates a number of repeated patterns, which appear tooccur between numerous different observers. Firstly, LowerPalaeolithic sites appear in limited frequency within the centralcore of the desert, and are predominately found at the periphery,whereas Middle and Late Palaeolithic populations occupy bothareas. Middle Palaeolithic assemblages occur on the widest varietyof raw material types, while Late Palaeolithic technologies arefocused upon fine silicious materials, such as jasper, and LowerPalaeolithic tools are typically made on coarser materials such asquartzites. Significant typological differences occur between thethree industries as well: Lower Palaeolithic assemblages promi-nently feature various heavy duty tools with basic flake productionand limited retouching; Middle Palaeolithic assemblages include awide variety of retouched tool types and the production of flakesfrom Levallois and discoidal cores, while preserving the productionof diminutive handaxes; Late Palaeolithic assemblages are typicallyfocused upon blade production, and include the use of backedmicroliths as well as other retouched tool types.

Dated and excavated evidence for hominin behaviour in thePalaeolithic Thar Desert is restricted to two sites: 16R Dune (Misraand Rajaguru, 1986, 1989; Raghavan et al., 1989; Blinkhorn, 2013)and Katoati (Blinkhorn et al., 2013) (Fig. 1). Middle Palaeolithicassemblages from Katoati featuring Levallois production as well asretouching of points date between 100 and 60 ka, overlapping aMiddle Palaeolithic occupation at 16R Dune bracketed by dates of80e40 ka. There is no clear basis upon which to assign an earlieroccupation at 16R Dune, dating between 130 and 109 ka, to eitherthe Middle or Lower Palaeolithic. In the Orsang Valley, Ajithprasad(2005) has identified Late Acheulean artefacts in sediments datedby Juyal et al. (2003) to 125e97 ka andMiddle Palaeolithic artefactsin sediments dating to 63e49 ka, presenting an earlier and suc-cessful example survey of dated sediments. Baskaran et al. (1986)identified Lower Palaeolithic artefacts in gravels at Junagadhdating to >69 ka and Middle Palaeolithic artefacts in gravels datingto >56 ka at Jetpur, while Mishra et al. (1999) report a small andundiagnostic collection of artefacts at Karna dating to 80 ka. Theearliest reported Late Palaeolithic artefacts come from an MIS 2context near Katoati.

From this brief overview, it is apparent that despite offering awealth of opportunity, the Palaeolithic record of the Thar Desert isyet to mature. The refinement of technological, rather than purelytypological, differences between the major Palaeolithic phases iscertainly a key endeavour, both in the Thar and across South Asia toenable global comparisons. Yet typically, such issues are bestaddressed through the study of large, systematically excavatedassemblages, requiring significant resources and investment. Incontrast, given the numerous sediment formations identified anddated by palaeoenvironmental researchers, surface survey offersthe potential to constrain the chronology of Palaeolithic occupationof the desert and patterns of typological change.

rface sites occurring on dated sediment formations in the Thar Desert,1.027

J. Blinkhorn / Quaternary International xxx (2014) 1e114

4. Survey sites

Three sites were selected for survey in central Rajasthan basedupon the presence of well dated sediment formations: Chamu,Karna and Shergarh Tri-Junction (see Fig. 1). A small collection ofartefacts has previously been described from the dated deposits atKarna (Mishra et al., 1999), indicating its suitability for furthersurvey. Reports of bothMiddle and Late Palaeolithic sites have beenmade from sites near to Shergarh Tri-Junction prior to palae-oenvironmental research and chronometric dating of sediments atthis location. In contrast, there are no previous reports of Palae-olithic surface sites in the vicinity of Chamu, although the sedimentformation is analogous to that of 16R Dune.

4.1. Chamu

The construction of an irrigation canal near the village of Chamuresulted in substantial excavations through a large dune deposit,revealing laterally extensive and deep sedimentary sections. Dhiret al. (2010) have produced a composite stratigraphy for the sedi-mentary sequence exposed near Chamu that is dated by OSL(Fig. 2a). This sequence comprises seven units, commencing withcemented channel deposits that are not dated, overlain by acompact aeolian horizon dated to 155 � 31 ka and a thin, verycompact, undated aeolian deposit (Dhir et al., 2010). This is overlainby Unit IV, a thick friable aeolian sand with ground water relatedlensoid features and carbonate bands in the lower portion andpatchy nodular calcretes in the upper third (Dhir et al., 2010). SevenOSL dates are reported between 105 � 3.8 ka and 81 � 11 ka, andalthough they do not occur in stratigraphic order this horizon canbe dated to MIS 5c. Unit V consists of a pedogenised aeolian sandwith common, hard calcrete nodules, dating to 58 � 5.1 ka and isoverlain by loose, weakly pedogenised aeolian sands dating to 27 kain the lower portion and 17 ka in the upper portion of Unit VI. The

Fig. 2. Sediment profiles and associated OSL dates (ka) from: a) Chamu (adapted from Dh(adapted from Andrews et al. (1998)).

Please cite this article in press as: Blinkhorn, J., Late Middle Palaeolithic suQuaternary International (2014), http://dx.doi.org/10.1016/j.quaint.2014.0

sequence is capped by a friable recent aeolian sand. The results ofstable carbon isotope analysis from Units IV, V, and VI range be-tween �2.03& and 0.51&, suggesting limited differences in vege-tation at the site, which would have been dominated by C4 planttypes. Oxygen isotope analyses indicate calcretes in Units IV and Vformed from slight to moderately evaporated monsoon precipita-tion, ranging between �6.8& and �4.0&, whereas results fromUnit VI suggest more intensive evaporation. Dhir et al. (2010)suggest that dune aggradation was episodic, followed by periodsof climatic amelioration permitting calcrete formation.

4.2. Karna

Karna is a cliff section of fluvial deposits situated in the LuniValley, which is the only existing major fluvial feature to draincentral Rajasthan. Currently, the river is seasonal, flowing princi-pally during and directly after the monsoon period, after which thesurface waters fragment into isolated ponds, although a period ofdeep incision, apparently occurring in the Holocene wet phasepresents large cliff sections in a number of areas in which the LatePleistocene behaviour of the Luni can be discerned. In a study of sixsections from the Mid-Luni Valley, Jain et al. (2005) have definedtwo distinct alluvial successions. The first (Type 1) consists of anextensive suite of vertically stacked, upward fining cycles depositedby gravelly braided streams that progress to form floodplain facies,but these were beyond the age range of the applied OSL techniquesto >300 ka (Jain et al., 2005). While the occurrence of cementedgravels at the base of the sequence from Chamu may not directlyrelate to the Luni drainage system, it occurs underlying a date of155 ka, offering broad support for fluvial activity during the Mid-Pleistocene (Dhir et al., 2010).

The second sequence (Type 2) illustrates depositional environ-ments dating fromMIS 5-1 (Fig. 2b). Humid phases are evidenced atKarna during MIS 5 by the deposition of multistoried gravel sheets

ir et al. (2010)); b) Karna (adapted from Jain et al. (2005)); and Shergarh Tri-Junction

rface sites occurring on dated sediment formations in the Thar Desert,1.027

J. Blinkhorn / Quaternary International xxx (2014) 1e11 5

associated with overbank calcic verstisols, interpreted as havingbeen deposited by gravel bed braided stream channels and adjacentpedogenised floodplains (Jain et al., 2005). These gravel depositsare dated by OSL to 79.8 � 8 ka. Overlying these gravels are hori-zontally bedded sands, dating between 29.7 � 2.8 ka and22.5 � 2.2 ka, indicating the presence of ephemeral sand bedstreams. A gravel bed dating to 11.4 � 1.2 ka caps the Pleistocenesequence at Karna, while mud and sand dominated horizons havebeen deposited in the Holocene. Combined with results fromelsewhere in the Mid Luni Valley, these results indicate a period ofstrong fluvial activity in MIS 5, followed by pedogenesis and sheetfloods in MIS 4 and 3 with ephemeral fluvial and aeolian activityoccurring with the transition to MIS 2.

4.3. Shergarh Tri-Junction

The study of Shergarh Tri-Junction (Andrews et al., 1998) hasbeen critical in the development of research into isotopegeochemistry of Upper Pleistocene sediment deposits from theThar Desert. A 4 m deep section in an abandoned gravel quarry wasstudied, revealing an alternating sequence of colluvial and aeoliandeposits, in the rocky pediment at the base of a prominent rhyolitehill (Fig. 2c) (Andrews et al., 1998). Five extensive aeolian units arereported, along with an impersistent aeolian horizon in the up-permost colluvial deposit. The four lower aeolian units (IeIV)contain calcrete nodules, whereas the uppermost unit (V) is nodulefree. The colluvial deposits consist of angular rhyolite gravels be-tween 2 and 5 cm diameter that are weakly cemented by calciumcarbonate in places. From the base upwards, aeolian units are datedto: Unit I e 60� 10 ka; Unit II e 64� 6 ka; Unit IIIe 55� 6 ka; UnitIV e 43 � 5 ka; and Unit V e 9 � 1 ka. Carbon isotope variation islimited, ranging between 0& and �1.7& across all aeolian units,which is indicative of dominantly C4 vegetation types. Greatervariation is observed in oxygen isotope results. The d18O resultsfrom Units II, III and the lower part of IV range between �4.5& and�6.4&, whereas results from I, the upper part of IV and theimpersistent aeolian unit range from �4.6& to �2.1&. Andrewset al. (1998) suggest positive covariation in the isotope resultsmay indicate evidence for stronger monsoonal rainfall from car-bonates in Unit II, III and the lower part of IV, although a number ofalternative interpretations are possible.

Table 1Basic lithic artefact typology and associated age of sites in central Rajasthan.

CHA1 CHA2 KAR1

Associated age 58e27 ka 105e58 ka 79.8 k

CoreSingle platform e 1 e

Multi platform e 2 1Bi-directional e e e

Prepared e e e

Levallois e e e

DebitageFlake 20 28 4Broken flake 18 9 1Core management flake 1 1 e

Flaked piece 9 3 1RetouchedSingle side e 2 e

Single end 1 e e

Single end single side 1 3 2Double side e e e

Single end double side e 1 1

Total 50 50 10

Please cite this article in press as: Blinkhorn, J., Late Middle Palaeolithic suQuaternary International (2014), http://dx.doi.org/10.1016/j.quaint.2014.0

5. Methods

Survey of the three sites was undertaken on foot. Preliminarysurvey was undertaken in order to identify the sediment sequencesreported by previous researchers. Following this, targeted archae-ological prospectionwas conducted in exposures of dated sedimentformations to discover lithic artefacts. All lithic analyses wereconducted in the field, using digital calipers and scales, aiming tocause minimal displacement to surface scatters of artefacts. Thesample size of artefact assemblages recorded was necessarilylimited by time available to conduct both prospection and analysis,and varies from 10 to 50 pieces. A range of cores, flakes andretouched pieces were selected in order to characterise the range ofvariability of the assemblages, as it was not possible to conduct asystematic assessment in the time available. Fresh artefacts lackingsignificant evidence forweathering or roundingwere targeted in anattempt to limit potential inclusion of transported artefacts. Thelithic attributes recorded include weight, maximum and axial sizevariables, raw material type, platform type, retouch location, andflake scar patterns. A brief description of the assemblage contextfollows, detailing both the relationship with previously datedsediments and taphonomic insights. Given the limited sample sizeof the assemblages, basic typological and technological de-scriptions of the artefacts are then presented.

6. Results

6.1. Assemblage provenance

A total of seven assemblages have been recorded from surfacesites associated with previously dated sediment formations. A basictypological description of the artefact assemblages is presented inTable 1. Two sites are reported from Chamu. The first (CHA1) wasrecovered from the surface of Horizon V, suggesting the assemblagepost-dates the age of 58 ka attributed to the horizon. It is likely thatthis assemblage pre-dates the formation of the friable, weaklypedogenised deposits of Horizon VI ca. 27 ka, although the rela-tionship between these two horizons was not directly evident.

The context of the CHA 2 assemblage is more problematic,occurring as a dump of material on top of Horizon V. The sur-rounding area is comprised of Horizon V sediments, with minimal

KAR2 KAR3 TRI1 TRI2

a 79.8 ka 79.8 ka 60e43 ka <43 ka

3 e 5 e

e 1 8 7e e 3 e

1 e 2 32 1 6 4

8 15 14 43 1 4 1

e 1 1 e

2 e 2 1

e e 3 e

e e 1 e

e 1 e e

e e 1 e

1 e e e

20 20 50 20

rface sites occurring on dated sediment formations in the Thar Desert,1.027

J. Blinkhorn / Quaternary International xxx (2014) 1e116

amounts of recent (Horizon VII) aeolian accumulation. The bank ofthe canal has been formed from the deposition of excavated ma-terial from the canal channel. As a result, it is suggested that thesandstone accumulations are dumps of material from the excava-tion of the canal channel. As Horizon V is the lowest observabledeposit in this location, this is the youngest deposit fromwhich thismaterial may derive. It is also plausible that the material derivesfrom the upper levels of Horizon IV. As a result, the CHA 2assemblage can be constrained by the dates of these two horizons,providing an age bracket of 58 ka to 81e105 ka.

During the survey at Karna, two lithic scatters (KAR1 & KAR2)were encountered close to one another near the current course of theLuni River. A small sample of 10 artefacts was recovered at KAR1 atthe eroding interface between a sandy gravel and an underlyingpedogenised sand deposit, displaying well developed pedogeniccarbonate nodules. At KAR2, a number of rhyolite boulders thatdisplay large,weatheredflake scars occuron topof a cementedgravelhorizon, and a sample of 20 artefacts have been analysed from theimmediate vicinity. A discrete lithic scatter (KAR3) was identified ontop of the calcic vertisol and a sample of 20 artefacts were analysed.The presence of numerous very small lithic fragments (<10mm), thefresh appearance of the artefacts and their occurrence as a spatiallydiscrete cluster suggest minimal natural disturbance. Given thestratigraphic position of all three assemblages, it is possible to sug-gest that they post-date themulti-storeyed gravel sheets in the area,which date to 79.8 ka (Jain et al., 2005), however, it is not possible tooffer a terminus ante quem date for these samples.

Two lithic assemblages have been recorded at Shergarh Tri-Junction. The first (TRI1) comprises 50 artefacts and was recov-ered from the quarry floor and predominately recovered fromdumps of pedogenic carbonates that are found at the edge of thequarry. On the basis of the presence of large pedogenic carbonatesnodules and the paucity of gravel content, this assemblage is likelyto derive from the dated aeolian horizons at the site, which havebeen deposited between 60 and 43 ka. The second assemblage(TRI2) is smaller, comprising of 20 artefacts. This assemblage wassampled from a spatially discrete surface scatter located on top ofthe upper-most colluvial unit. Occurring with a number of pedo-genic carbonates and rhizo-concretions on an exposed surface, it ispossible that this assemblage has been deflated. However, as theassemblage appears spatially confined and with similar levels ofartefact weathering, there does not appear to have been substantialmixing. As a result, the suggested age of this assemblage is <43 ka.

6.2. Raw materials

Raw material variability at all seven sites is limited. All but oneartefact from Shergarh Tri-Junction, and all but four from Karna areproduced on rhyolites, whereas all artefacts from Chamu are pro-duced on sandstones. A single sandstone artefact was identified atTRI1, whereas two quartzite and two chert lithics were recorded atKAR1. Rhyolite is abundantly available at Shergarh Tri-Junction inboth small clast sizes, as apparent in the local gravels, and largercobbles, boulders and outcrops from the adjacent hill. At Karna,rhyolite is available in the form of large boulders that appear on topof the cemented gravel horizon, which bear large flake scars,whereas larger outcrops are visible from the site ca. 5 km away. Incontrast, there are no sources for quartzite or chert in the imme-diate vicinity, which may either suggest these artefacts were im-ported to the site or suitable clasts for reduction were recoveredfrom the now cemented gravel horizons. No rawmaterial resourcesare evident in the Chamu landscape today, as the area is extensivelycovered in sand dunes. However, the overall character of the as-semblages would not suggest that they have been transported farfrom a raw material source (see below) suggesting a local source of

Please cite this article in press as: Blinkhorn, J., Late Middle Palaeolithic suQuaternary International (2014), http://dx.doi.org/10.1016/j.quaint.2014.0

sandstone may have been present in the Upper Pleistocene that hasnow been buried. No cortical pieces were recorded at Shergarh Tri-Junction, and only two artefacts in each assemblage from Chamupreserve some cortex. In contrast, five artefacts from KAR2 andseven from KAR3 preserve cortical surfaces indicating some pri-mary reduction activities occurred at these sites.

6.3. Cores

The majority of cores identified occur in the TRI1 and TRI2 as-semblages, although only CHA1 lacks cores completely. The threesites at Karna and CHA2 include low numbers of informal singleand multi-platform cores, with a higher frequency of such informalcores also appearing in the Shergarh Tri-Junction assemblages. BothLevallois and prepared cores (presenting some but not all techno-logical features commonly used to define Levallois cores) appear inKAR2 and both Shergarh Tri-junction sites, whereas a Levallois corewas also identified at KAR3.

Two of the three bidirectional cores from TRI1 exhibit a numberof parallel, longitudinal arises indicating that these have beenprepared as blade cores (Fig. 3a and b). The third bidirectional coreexhibits a single pair of parallel arises resulting from a final flakeremoval of a blade proportioned flake (Fig. 3c), although this coredoes not appear to have been as competently prepared for repeatedblade removals. Two Levallois cores from TRI2 also exhibit bidi-rectional removal of blade-proportioned flakes creating repeatedparallel arises (Fig. 3d and f), with a further bidirectional Levalloiscore presenting a single parallel sided blade scar as the finalremoval (Fig. 3e). A Levallois core from TRI1 presents a single bladeremoval, with the lateral shaping of the flaking surface likely tohave created a central crested ridge (Fig. 4g). Elsewhere, a singleprepared core (TRI2), single platform core (KAR2) and multi-platform core (TRI1) exhibit blade proportioned final flake scarsfrom less formalised flaking strategies.

Eight Levallois cores, including those from KAR2 and KAR3,exhibit preferential removals from centripetally flaked cores(Fig. 3hep). The majority of prepared cores also present a flakingsurface with a radial or weakly radial flaking pattern, suggestingthat they could represent preparatory phases of Levallois reduc-tion schemes that were either abandoned before completion orproduced suitable blanks without fulfilling the criteria to be clas-sified as Levallois cores (Fig. 3qes). A single triangular shapedLevallois core from TRI1 exhibits numerous distal-lateral removalscreating a distinct medial ridge, combined with unidirectionalshaping of the proximal flaking surface (Fig. 3t). The final removalfrom the core is on alignment with the distal ridge, but has steppedprior to reaching the laterally shaped distal portion of the core.This pattern of core reduction is analogous to the Nubian Type 2method.

The gross dimensions of the cores typically range between 40e100 m length and 30e70 mmwidth, with the single disc core fromKAR2 and individual multi platform cores from CHA2 and KAR3exceeding these dimensions. Within this range, only the threebidirectional cores exhibit a distinct restriction in size variability(all less than 55 � 44 mm). Complex flaking platforms (dihedral,multiple conchoidal or facetted types) are commonplace for theLevallois and prepared cores, but regularly feature on other coretypes as well as simple forms (plain and single conchoidal types).The majority of cores presented feather terminations, with threemulti-platform cores and a single Levallois core having beenidentified with stepped terminations on the final flake removal.Overall, cores do not appear to have been heavily rotated, with asingle instance of a multi-platform core from TRI1 exhibiting threerotations of flaking direction, suggesting few cores had beenexhausted before discard.

rface sites occurring on dated sediment formations in the Thar Desert,1.027

Fig. 3. Cores from sites reported in this study: aec) Bidirectional cores; dep) Levallois cores; qes) Prepared Cores; t) Levallois core.

J. Blinkhorn / Quaternary International xxx (2014) 1e11 7

6.4. Debitage

Debitage classes are dominated by the sites from Chamu, whichis likely to be an artefact of recovery methods and small samplessizes. Minimal differences in the range of gross size of debitage existbetween rhyolite and sandstone artefacts. However, a greaterproportion of larger debitage occurs on sandstone, suggesting raw

Please cite this article in press as: Blinkhorn, J., Late Middle Palaeolithic suQuaternary International (2014), http://dx.doi.org/10.1016/j.quaint.2014.0

material effects may have affecting reduction strategies and tech-niques, although sampling methodology and sample size may alsohave some effect. However, rhyolite artefacts from both ShergarhTri-Junction and Karna typically range between 20e70 mm lengthand 10e40 mm width, Sandstone artefacts regularly occur up to100 mm � 80 mm, while also overlapping heavily with the smallerrange exhibited by rhyolite. Although flakes, broken flakes and

rface sites occurring on dated sediment formations in the Thar Desert,1.027

Fig. 4. Retouched artefacts from sites reported in this study: CHA1 e a & l; CHA2 e b, d, e, g & h; KAR1 e f, i & m; KAR2 e j; KAR3 e k; TRI1 e n.

J. Blinkhorn / Quaternary International xxx (2014) 1e118

flaked pieces comprise the majority of the unretouched debitageassemblages, individual core management flakes were present atCHA1, CHA2, KAR3 and TRI1. The presence of these artefacts sug-gests some degree of preparation and control over flaking strate-gies, rather than more simplistic or expedient reduction methods,which in the case of Chamuwere not evident in the low numbers ofcores seen.

The majority of complete flakes occur within the range of 30e100 mm in length and 30e80 mm in width. Blade proportionedcomplete flakes (n ¼ 12) occur in low proportions at all sitesexcept for KAR1, with L:W ratios ranging between 2e2.85. Com-plete flakes with sub-blade proportions (L:W ratio > 1.5) occurmore frequently (n ¼ 29) at all sites, including an example withparallel arises on the dorsal surface. However, the majority ofcomplete flakes (n ¼ 49) exhibit more squared proportions. Flakeswith proximal scars are most commonplace (n ¼ 29), whereasother forms of linear reduction, indicated by flakes with distal(n ¼ 2) or bidirectional (n ¼ 1) dorsal scar patterns, are highlylimited. Flakes exhibiting lateral or perpendicular scar patterns(n ¼ 21) also appear regularly in all sites. Four of six radial flakeswere recovered from TRI1, with the other two coming fromChamu, while eight of thirteen weakly radial flakes were alsorecovered from Chamu, with a further three weakly radial piecesoccurring at TRI1.

6.5. Retouched artefacts

Retouched artefacts were recovered in very low numbers (totaln ¼ 18), but examples were identified in all assemblages except for

Please cite this article in press as: Blinkhorn, J., Late Middle Palaeolithic suQuaternary International (2014), http://dx.doi.org/10.1016/j.quaint.2014.0

TRI2. The most common locations for retouching occur on either asingle side (n ¼ 6), or extending from a single end to a single side(n ¼ 6). Three artefacts exhibit retouch extending across both sidesand an end, with two pieces presenting retouch only on a singleend, and one piece with retouch on both sides but neither end.

Two retouched artefacts were identified in CHA1. A broad flakewith a facetted platform has been delineated on the left and distalmargins of the dorsal face, including the creation of a smallshoulder on the side of the piece (Fig. 4a). Thinning of this side isobserved by more invasive flaking on the ventral surface, whilefurther thinning of the artefact is apparent on the opposing,tapering side. Bifacial delineation of the side and distal margin of asmall piece has created a symmetrical outline, including a smallprotrusion opposed to its tapering but snapped edge (Fig. 4l).

CHA2 includes the greatest number of retouched pieces. Themost notable retouched artefact is a bifacially flaked foliate stem-med point (Fig. 4h). The stem is formed through a notch and a largeinvasive removal, while the outline of the piece has been formedthrough non-invasive retouch of both edges. Two divergent re-movals help to form the point that opposes the stemmed end of theartefact. A flake with a medial break exhibits retouch solely on thedorsal face on a single end and side, significantly narrowing theprofile of the artefact to form a stem or incipient tang (Fig. 4d). Asecond piece with a medial break also has a retouched end, withminor retouching accomplishing a narrowed and projected outlineopposing the break (Fig. 4e). A core management flake provides afurther example of where minor retouching has been employed toestablish a narrow protrusion on one side of the artefact (Fig. 4b).Retouch along on a single side of the dorsal face of a large flake has

rface sites occurring on dated sediment formations in the Thar Desert,1.027

J. Blinkhorn / Quaternary International xxx (2014) 1e11 9

formed a pointed outline (Fig. 4c). A blade with a snapped endexhibits informal retouch on a single side (Fig. 4g).

A flake from KAR1 exhibits invasive retouching of the distalventral surface and a protruding outline is created on the left side ofthe piece by two simple notches, the smaller of which is reinforcedwith further removals on the dorsal surface (Fig. 4m). A secondflake shows a delineated side on the ventral face, extending ontothe longer distal edge, creating a shouldered outline, with atapering outline on the opposite side that has been snapped,potentially through use (Fig. 4i). Amoreweathered artefact exhibitsa stem on an end, formed by small retouch removals accentuating alarger removal, and further trimming is observed on one side(Fig. 4f). This piece appears to have been substantially larger priorto a significant medial break. Numerous small removals on bothsurfaces of the end of a flaked piece from KAR2 create two smallopposed notches and create a thin protrusion that may be classifiedas either a stem or tang (Fig. 4j). The retouched artefact from KAR3presents an invasive removal on the left dorsal surface, with minorremovals accentuating a tapering projection on the same side buton the ventral face, with less invasive trimming on the dorsal distaledge (Fig. 4k). Unlike the core population, retouched artefacts fromTRI1 present few diagnostic features. An artefact with notchesformed on either side of a flake was identified (Fig. 4n), with theremaining pieces informally retouched either on a single end(n ¼ 2) or side (n ¼ 2).

6.6. Summary

The artefact assemblages and their sedimentary context re-ported here offer significant, albeit tentative, insights into the na-ture and chronology of lithic typo-technology from the UpperPleistocene in central Rajasthan. These sites occur further towardthe core of the Thar Desert than other sites associated with chro-nometric dates discussed above, and therefore are important toconsider the timing of inhabitation of the region and culturalpractices associated with Palaeolithic occupations. Due to thechronology of the sedimentary contexts targeted by this surveywork, spanning 100 ka to less than 43 ka, the nature of the lithicassemblages are informative for considering the impact of expan-sions of Homo sapiens into the Indian subcontinent, which arehypothesised to occur within this timespan (Boivin et al., 2013;Mellars et al., 2013; Mishra et al., 2013).

The oldest assemblage recorded is likely to have been CHA2,derived from an MIS 5c-4 sedimentary context, for which the mostdiagnostic feature of the lithic technology is the retouched toolassemblage. The bifacially flaked and stemmed foliate point offersthe clearest indication that retouched tools were likely beingmodified for hafting. This supports the suggestion that other formsof retouching observed at the site that create distinct, narrowprojections may also have been undertaken to integrate stone ar-tefacts into composite tools. The presence of two such tools wheresuch projections appear opposite medial breaks may indicate theirdiscard when they were no longer useful as hafted implements.Further examples of features associated with hafted technologies,such as tangs, stems, shoulders and basal thinning, can be observedin the younger site of CHA1, and in the three assemblages fromKarna. The presence for hafting of stone implements in the centralThar is supported by similar evidence identified in the easternmargin of the desert at the excavated and dated site of Katoati,where analogous retouched artefacts are also found spanningMIS 5e 4/3 (Blinkhorn et al., 2013).

Beyond the informal reduction techniques observed, coretechnology in the central Thar indicates the preferential productionof a number of different blank types. Most commonplace is theappearance of preferential Levallois flake production, identified in

Please cite this article in press as: Blinkhorn, J., Late Middle Palaeolithic suQuaternary International (2014), http://dx.doi.org/10.1016/j.quaint.2014.0

TRI1, TRI2, KAR2 and KAR3, suggesting the use of Levallois tech-nologies continued well into MIS 3 (<43 ka at TRI2). Prepared coresmay also have offered an alternative method for the production ofbroad radial flakes, or may represent steps on a common techno-logical trajectory with Levallois cores for the production of suchblanks.

Different approaches to blade production are evident at Sher-garh Tri-Junction. Both bidirectional cores and bidirectional Leval-lois cores have been exploited to produce repeated, parallel sidedblades, although some variability in the proficiency of this exploi-tation is apparent. This offers the earliest evidence for the devel-opment of formalised blade production strategies in the TharDesert, which appears to occur from the onset of MIS 3. Given thatLevallois reduction methods occur in the Thar Desert from at leastca. 96 ka, the overlap in bidirectional reduction techniques identi-fied here may offer critical evidence to suggest that blade produc-tion developed in this region from Levallois methods.

Two further Levallois cores offer insight into the variability ofpreferential flake production in central Rajasthan. A preferentialcrested blade removal evident on a Levallois core offers a differentapproach to the production of elongate blanks in the region. Somesimilarity in the use of lateral shaping to determine flake removalscan be observed when this is compared with the unsuccessfulremoval of a flake from a core where the shaping of the flakingsurface is analogous to the Nubian Type 2 method. Differencesbetween the two can be noted in how the distal convexity of theflaking surface has been managed and the impact of this on suc-cessful removal of preferential flakes. Their occurrence in TRI1suggests a chronological overlap with similar reduction methodsemployed at Katoati.

7. Discussion

The survey of previously dated sediment formations for surfacePalaeolithic sites offers a rapid means to assess the nature of lithictechnology capitalising upon robust terminus post quem dating inregions lacking excavated and dated archaeological sites. As aresult, this approach is most likely to be of use to offer preliminaryanswers to broad questions for which terminus post quem datingoffers suitable chronological resolution. In this case, the identifi-cation of lithic assemblages that overlie sediment horizons datingto the onset of MIS 3, such as TRI1, TRI2 and CHA1, are critical toilluminate the nature of stone tool technology at a time when fossiland genetic evidence clearly indicate that H. sapiens populationshave expanded from Africa to Australia. This is particularly perti-nent in the Thar Desert where the youngest clearly dated MiddlePalaeolithic assemblages appear at the MIS 4/3 boundary while theoldest Late Palaeolithic assemblages currently reported occur insediments post-dating the Last Glacial Maximum. Therefore, theresults presented here play a critical role in understanding thetransition from Middle to Late Palaeolithic technologies, and canhelp guide future excavation strategies.

The methodology employed in this study has been most suc-cessful in identifying diagnostic typo-technological features ofcores and retouched artefacts. In particular, the range of Levalloiscores and retouched tool types observed in all assemblages appearcongruous with Middle Palaeolithic assemblages identified bothwithin the Thar Desert and across South Asia (Clarkson et al., 2012;Haslam et al., 2012; Blinkhorn, 2013; Blinkhorn et al., 2013) andpotentially comparable with lithic industries observed further westin the mid-latitude arid belt (e.g. Groucutt and Petraglia, 2012; Roseet al., 2011; Usik et al., 2013; Scerri, 2013a, 2013b). Due to thelimitations of sample size, the analyses of unretouched debitagehave proved less diagnostic in this instance, although furtherinsight into technological changes may be possible when they are

rface sites occurring on dated sediment formations in the Thar Desert,1.027

J. Blinkhorn / Quaternary International xxx (2014) 1e1110

compared with Late Palaeolithic assemblages. The results of thissurvey, therefore, indicate the continued use of Middle Palaeolithictechnologies in the central Thar Desert into MIS 3, and post-dating43 ka. Significantly, the youngest assemblages reported offer someinsight into the development of blade technologies that appearrooted in Levallois core reduction schemes and may thus beindicative of localised developments of blade production. Thesepreliminary conclusions present hypotheses that can be testedthrough excavation and more direct dating of lithic assemblages.

Refining the chronology of late Middle Palaeolithic and earlyLate Palaeolithic industries is critical to understanding the nature ofthe transition between the two, which currently plays a central rolein the debate regarding the arrival of H. sapiens in South Asia. Twocontemporary models propose that Late Palaeolithic industries,typically identified as microblade industries accompanied bybacked artefacts, were not developed in South Asia but permittedmodern humans to rapidly colonise the coastline (Mellars et al.,2013) or readily outcompete archaic human populations usingMiddle Palaeolithic technologies to colonise the subcontinents’interior (Mishra et al., 2013). An alternative model suggests modernhumans colonised South Asia using Middle Palaeolithic technolo-gies during MIS 5, and Late Palaeolithic technologies were subse-quently developed in response to local environmental,demographic and cultural factors (Boivin et al., 2013). Discerning alocal or non-local origin of blade/microblade and backed technol-ogies and the timing of the appearance of these features withinlithic assemblages are crucial to evaluate these models.

The appearance of blade technologies rooted in bidirectionalLevallois reduction strategies at the onset of MIS 3 at Shergarh Tri-Junction suggests aspects of continuity between Middle and LatePalaeolithic technologies. Notably, this occurs at a time whenpalaeoenvironmental records indicate the onset of fluctuating hu-midity within the Thar Desert, which may have offered suitablecontexts for demographic expansions. The development of bladetechnologies in the Thar Desert could relate to larger effectivepopulation sizes (e.g. Atkinson et al., 2008) permitted by the returnof humid conditions in the region, enabling adaptive technologicalinnovations. Furthermore, the identification of a core analogous toNubian Type 2 cores found in the Middle Palaeolithic of North-EastAfrica and Arabia compounds evidence from Katoati to suggestaspects of cultural continuity in Middle Palaeolithic assemblagesacross the mid-latitude arid belt. Significant overlap between theassociated chronology of the youngest Middle Palaeolithic assem-blages reported here and both fossil and genetic evidence for theexistence of modern human populations in South-East Asia andAustralia occurs, such that a modern human authorship for MiddlePalaeolithic technologies in the Thar Desert cannot be precluded.The findings of this study suggest a local development of systematicblade production as well as aspects of cultural continuity across themid-latitude arid belt, which support models of human dispersalthat indicate Middle Palaeolithic technologies were indeed suffi-cient to colonise dramatically new environments.

8. Conclusion

This research highlights the significance of late Middle Palae-olithic assemblages in the context of contemporary debateregarding the dispersals of modern humans into South Asia. Byformulating a suitable research problem that can be addressed withsupport from terminus post-quem dating, it has been possible tointegrate archaeological evidence from surface sites occurring ondated sediment formations into a broader framework of environ-mental change and cultural evolution in themid Upper Pleistocene.The results of the survey present the first dated evidence for UpperPleistocene occupations in the central Thar, indicating the

Please cite this article in press as: Blinkhorn, J., Late Middle Palaeolithic suQuaternary International (2014), http://dx.doi.org/10.1016/j.quaint.2014.0

occurrence of lithic assemblages in MIS 3 contexts that presentdiagnostic features permitting their attribution to Middle Palae-olithic industries. Importantly, these assemblages offer insightsinto the regional trajectory of technological development fromMiddle to Late Palaeolithic industries in the Thar Desert in thecontext of climatic amelioration and population expansions.

Acknowledgements

Prof. Hema Achyuthan, Prof. Michael D. Petraglia and Dr. NicoleBoivin are sincerely thanked for their support of the conduct of thisresearch. The Archaeological Survey of India is acknowledged forlogistical support. This research has been supported through theEmslie Horniman Scholarship (Royal Anthropological Institute), theKathleen Hardy Scholarship (St. Hugh’s College, University of Ox-ford) and a Fondation Fyssen Post-Doctoral Fellowship. I thank theeditors, Huw Groucutt and Eleanor Scerri, for the invitation tocontribute to this special issue.

References

Achyuthan, H., Quade, J., Roe, L., Placzek, C., 2007. Stable isotopic composition ofpedogenic carbonates from the eastern margin of the Thar Desert, Rajasthan,India. Quaternary International 162e163, 50e60.

Ajithprasad, P., 2005. Early middle Palaeolithic: a transition phase between theupper Acheulian and Middle Palaeolithic cultures in the Orsang Valley, Gujarat.Man and Environment 30, 1e11.

Allchin, B., Goudie, A., Hegde, K.T.M., 1978. The Prehistory and Palaeogeography ofthe Great Indian Desert. Academic Press, London.

An, Z., Clemens, S.C., Shen, J., Qiang, X., Jin, Z., Sun, Y., Prell, W.L., Luo, J., Wang, S.,Xu, H., Cai, Y., Zhou, W., Liu, X., Liu, W., Shi, Z., Yan, L., Xiao, X., Chang, H., Wu, F.,Ai, L., Lu, F., 2011. GlacialeInterglacial Indian summer monsoon dynamics.Science 333, 719e723.

Andrews, J.E., Singhvi, A.K., Kailath, A.J., Kuhn, R., Dennis, P.F., Tandon, S.K., Dhir, R.P.,1998. Do stable isotope data from calcrete record Late Pleistocene monsoonalclimate variation in the Thar Desert of India? Quaternary Research 50, 240e251.

Arzarello, M., Boudad, L., Guislain, S., 2013. Middle Paleolithic occupation of theMoroccan Sahara: open air sites of the Tafilalt. Quaternary International 300,131e141.

Atkinson, Q.D., Gray, R.D., Drummond, A.J., 2008. mtDNA variation predicts popu-lation size in humans and reveals a major Southern Asian chapter in humanprehistory. Molecular Biology and Evolution 25, 468e474.

Bajpai, V.N., Saha Roy, T.K., Tandon, S.K., 2001. Subsurface sediment accumulationpatterns and their relationships with tectonic lineaments in the semi-arid Luniriver basin, Rajasthan, Western India. Journal of Arid Environments 48, 603e621.

Baskaran, M., Maratheb, A.R., Rajagurub, S.N., Somayajulu, B.L.K., 1986. Geochro-nology of Palaeolithic cultures in the Hiran Valley, Saurashtra, India. Journal ofArchaeological Science 13, 505e514.

Beyin, A., 2013. A surface Middle Stone Age assemblage from the Red Sea coast ofEritrea: implications for Upper Pleistocene human dispersals out of Africa.Quaternary International 300, 195e212.

Blinkhorn, J., 2012. The Palaeolithic Occupation of the Thar Desert. D.Phil. Thesis.University of Oxford, Oxford.

Blinkhorn, J., 2013. A new synthesis of evidence for the Upper Pleistocene occu-pation of 16R Dune and its southern Asian context. Quaternary International300, 282e291.

Blinkhorn, J., Achyuthan, H., Petraglia, M., Ditchfield, P., 2013. Middle Palaeolithicoccupation in the Thar Desert during the Upper Pleistocene: the signature of amodern human exit out of Africa? Quaternary Science Reviews 77, 233e238.

Boivin, N., Fuller, D.Q., Dennell, R., Allaby, R., Petraglia, M.D., 2013. Human dispersalacross diverse environments of Asia during the Upper Pleistocene. QuaternaryInternational 300, 32e47.

Cancellieri, E., di Lernia, S., 2013. Middle Stone Age human occupation and dis-persals in the Messak plateau (SW Libya, central Sahara). Quaternary Interna-tional 300, 142e152.

Chawla, S., Dhir, R., Singhvi, A., 1992. Thermoluminescence chronology of sandprofiles in the Thar desert and their implications. Quaternary Science Reviews11, 25e32.

Clarkson, C., Jones, S., Harris, C., 2012. Continuity and change in the lithic industriesof the Jurreru Valley, India, before and after the Toba eruption. QuaternaryInternational 258, 165e179.

Delagnes, A., Crassard, R., Bertran, P., Sitzia, L., 2013. Cultural and human dynamicsin southern Arabia at the end of the Middle Paleolithic. Quaternary Interna-tional 300, 234e243.

Delagnes, A., Tribolo, C., Bertran, P., Brenet, M., Crassard, R., Jaubert, J., Khalidi, L.,Mercier, N., Nomade, S., Peigné, S., Sitzia, L., Tournepiche, J.-F., Al-Halibi, M., Al-Mosabi, A., Macchiarelli, R., 2012. Inland human settlement in southern Arabia

rface sites occurring on dated sediment formations in the Thar Desert,1.027

J. Blinkhorn / Quaternary International xxx (2014) 1e11 11

55,000 years ago. New evidence from the Wadi Surdud Middle Paleolithic sitecomplex, western Yemen. Journal of Human Evolution 63, 452e474.

Deotare, B.C., Kajale, M.D., Rajaguru, S.N., Kusumgar, S., Jull, A.J.T., Donahue, J.D.,2004. Palaeoenvironmental history of Bap-Malar and Kanod playas of westernRajasthan, Thar Desert. Journal of Earth System Science 113, 403e425.

Dhir, R.P., Singhvi, A.K., Andrews, J.E., Kar, A., Sareen, B.K., Tandon, S.K., Kailath, A.,Thomas, J.V., 2010. Multiple episodes of aggradation and calcrete formation inLate Quaternary aeolian sands, Central Thar Desert, Rajasthan, India. Journal ofAsian Earth Sciences 37, 10e16.

Enzel, Y., Ely, L., Mishra, S., Ramesh, R., Amit, R., Lazar, B., Rajaguru, S., Baker, V.,Sandler, A., 1999. High-resolution holocene environmental changes in the TharDesert, Northwestern India. Science 284, 125e128.

Finlayson, C., 2013. The water optimisation hypothesis and the human occupationof the mid-latitude belt in the Pleistocene. Quaternary International 300, 22e31.

Foley, R.A., Maíllo-Fernández, J.M., Mirazón Lahr, M., 2013. The Middle Stone Age ofthe Central Sahara: biogeographical opportunities and technological strategiesin later human evolution. Quaternary International 300, 153e170.

Groucutt, H.S., Blinkhorn, J., 2013. The Middle Palaeolithic in the desert and itsimplications for understanding hominin adaptation and dispersal. QuaternaryInternational 300, 1e12.

Groucutt, H.S., Petraglia, M.D., 2012. The prehistory of the Arabian peninsula: de-serts, dispersals, and demography. Evolutionary Anthropology 21, 113e125.

Haslam, M., Clarkson, C., Roberts, R.G., Bora, J., Korisettar, R., Ditchfield, P.,Chivas, A.R., Harris, C., Smith, V., Oh, A., Eksambekar, S., Boivin, N., Petraglia, M.,2012. A southern Indian Middle Palaeolithic occupation surface sealed by the 74ka Toba eruption: further evidence from Jwalapuram locality 22. QuaternaryInternational 258, 148e164.

Jain, M., Tandon, S., Singhvi, A., Mishra, S., Bhatt, S., 2005. Quaternary alluvialstratigraphical development in a desert setting: a case study from the LuniRiver basin, Thar Desert of western India. Fluvial Sedimentology VII, 349e371.

Juyal, N., Chamyal, L., Bhandari, S., Bhushan, R., Singhvi, A., 2006. Continentalrecord of the southwest monsoon during the last 130 ka: evidence from thesouthern margin of the Thar Desert, India. Quaternary Science Reviews 25,2632e2650.

Juyal, N., Kar, A., Rajaguru, S., Singhvi, A., 2003. Luminescence chronology of aeoliandeposition during the Late Quaternary on the southern margin of Thar Desert,India. Quaternary International 104, 87e98.

Liu, Z., 2011. Atmospheric science. Glacial cycles and Indian monsoonea southernpush. Science 333, 706e708.

Mellars, P., Gori, K.C., Carr, M., Soares, P.A., Richards, M.B., 2013. Genetic andarchaeological perspectives on the initial modern human colonization ofsouthern Asia. PNAS 110, 10699e10704.

Mishra, S., Chauhan, N., Singhvi, A.K., 2013. Continuity of microblade technology inthe Indian subcontinent since 45 ka: implications for the dispersal of modernhumans. PLoS ONE 8, e69280.

Mishra, S., Jain, M., Tandon, S.K., Singhvi, A.K., Joglekar, P.P., Bhatt, S.C.,Kshirsagar, A.A., Naik, S., 1999. Prehistoric cultures and Late Quaternary envi-ronments in the Luni Basin around Balotra. Man and Environment 24, 39e49.

Please cite this article in press as: Blinkhorn, J., Late Middle Palaeolithic suQuaternary International (2014), http://dx.doi.org/10.1016/j.quaint.2014.0

Misra, V.N., Rajaguru, S.N., Raju, D.R., Raghavan, H., Gaillard, C., 1982. Acheulianoccupation and evolving landscape around Didwana in the Thar Desert, India.Man and Environment 6, 72e86.

Misra, V.N., 1995. Geoarchaeology of the Thar Desert, North West India. In:Wadia, S., Korisettar, R., Kale, V. (Eds.), Quaternary Environments and Geo-archaeology of India. Geological Society of India, Bangalore, pp. 210e230.

Misra, V.N., Rajaguru, S.N., 1986. Environment et culture de l’Homme prehistoriquedans les desert du Thar, Rajasthan, Inde. L’Anthropologie 90, 407e437.

Misra, V.N., Rajaguru, S.N., 1989. Palaeoenvironments and prehistory of the TharDesert, Rajasthan, India. In: South Asian Archaeology 1985. ScandinavianInstitute of Asian Studies, Copenhagen, pp. 296e320. Occasional Papers.

Petraglia, M.D., Alsharekh, A.M., Crassard, R., Drake, N.A., Groucutt, H., Parker, A.G.,Roberts, R.G., 2011. Middle Paleolithic occupation on a Marine Isotope Stage 5lakeshore in the Nefud Desert, Saudi Arabia. Quaternary Science Reviews 30,1555e1559.

Petraglia, M.D., Groucutt, H., Blinkhorn, J., 2012. Hominin evolutionary history inthe Arabian Desert and the Thar Desert. In: Mol, S., Sternberg, T. (Eds.),Changing Deserts: Integrating People and Their Environment. White HorsePress, Cambridge, pp. 61e82.

Raghavan, H., Rajaguru, S.N., Misra, V.N., 1989. Radiometric dating of a Quaternarydune section, Didwana, Rajasthan. Man and Environment 13, 19e22.

Rose, J.I., Usik, V.I., Marks, A.E., Hilbert, Y.H., Galletti, C.S., Parton, A., Geiling, J.M.,Cerný, V., Morley, M.W., Roberts, R.G., 2011. The Nubian complex of Dhofar,Oman: an African Middle Stone Age industry in Southern Arabia. PloS One 6,e28239.

Scerri, E.M.L., 2013a. On the spatial and technological organisation of haftingmodifications in the North African Middle Stone Age. Journal of ArchaeologicalScience 40, 4234e4248.

Scerri, E.M.L., 2013b. The Aterian and its place in the North African Middle StoneAge. Quaternary International 300, 111e130.

Singhvi, A.K., Kar, A., 2004. The aeolian sedimentation record of the Thar desert.Proceedings of the Indian Academy of Sciences 113, 371e401.

Singhvi, A.K., Williams, M.A.J., Rajaguru, S.N., Misra, V.N., Chawla, S., Stokes, S.,Chauhan, N., Francis, T., Ganjoo, R.K., Humphreys, G.S., 2010. A w200 ka recordof climatic change and dune activity in the Thar Desert, India. QuaternaryScience Reviews 29, 3095e3105.

Sinha, R., Smykatz-Kloss, W., Stu, D., 2006. Late Quaternary palaeoclimatic recon-struction from the lacustrine sediments of the Sambhar playa core, Thar Desertmargin, India. Palaeogeography Palaeoclimatology Palaeoecology 233, 252e270.

Srivastava, P., Juyal, N., Singhvi, A.K., Wasson, R.J., Bateman, M.D., 2001. Lumines-cence chronology of river adjustment and incision of Quaternary sediments inthe alluvial plain of the Sabarmati River, north Gujarat, India. Geomorphology36, 217e229.

Usik, V.I., Rose, J.I., Hilbert, Y.H., Van Peer, P., Marks, A.E., 2013. Nubian complexreduction strategies in Dhofar, southern Oman. Quaternary International 300,244e266.

Zeuner, F.E., 1950. Stone Age and Pleistocene Chronology in Gujurat. Deccan College,Post-Graduate and Research Institute, Poona.

rface sites occurring on dated sediment formations in the Thar Desert,1.027