Hu Man Oc Cu Pa Tion, Ad Ap Ta Tion and Be Hav Ioral

Eurasian Pre history, 6 (1–2): 119–166.

HU MAN OC CU PA TION, AD AP TA TION AND BE HAV IORAL CHANGE IN THE PLEIS TOCENE AND HOLO CENE OF SOUTH IN DIA: RE CENT IN VES TI GA TIONS IN THE KURNOOL DIS TRICT, ANDHRA PRADESH

Mi chael Petraglia1, Ravi Korisettar2, M. Kasturi Bai3, Nicole Boivin1, Janardhana B.2, Chris to pher Clark son4, Kevin Cunningham5, Pe ter Ditchfield1, Dorian Fuller6, Jamie Hampson7, Michael Has lam1, Sacha Jones7, Jinu Koshy2, Pres ton Mir a cle7, Clive Oppenheimer8, Richard Roberts 9 and Kevin White10

1 Re search Labo ratory for Ar chae ol ogy and the His tory of Art, Dyson Perrins Build ing, South Parks Road, Univer sity of Ox ford, Ox ford, OX1 3QY, UK; mi [email protected], [email protected], pe [email protected], mi chael.has [email protected] 2 De partment of His tory and Ar chae ol ogy, Karnatak Uni ver sity, Dharwad, 580 003, In dia; [email protected], janard_97@ya hoo.co.in, jinu.knapper@ya hoo.com 3 c/o Ar chae olog i cal Sur vey of In dia, 8th Cross, Kalyan Nagar, Dharwad – 580 007, In dia 4 School of So cial Sci ence, The Univer sity of Queensland, St Lucia, Queensland 4072, Aus tralia; c.clark [email protected] 5 Del a ware De part ment of Trans por ta tion, Do ver, Del a ware, 19903, USA; [email protected] 6 Insti tute of Archae ol ogy, Uni ver sity Col lege Lon don, 31-34 Gordon Square, London, WC1H 0PY, UK; [email protected] 7 De partment of Ar chae ol ogy, Uni ver sity of Cam bridge, Downing Street, Cam bridge, CB2 3DZ, UK; [email protected], [email protected], [email protected] 8 De partment of Ge ogra phy, Uni ver sity of Cam bridge, Downing Place, Cambridge CB2 3EN, UK; [email protected] 9GeoQuEST Re search Cen tre, School of Earth and En vi ronmen tal Sci ences, Uni ver sity of Wollongong, Wollongong, New South Wales 252, Aus tralia; [email protected] 10 De partment of Ge ogra phy, The Uni ver sity of Reading, Whiteknights, Read ing, RG6 6AB, UK; k.h.white@read ing.ac.uk

Ab stract This arti cle reports on ongo ing archae o log i cal research in the Kurnool District of southern India. The region is well-known for its rich cultural heri tage, with previ ous inves ti ga tors identi fy ing “Upper Paleolithic” and “Mesolithic” in - dustries in caves, and Neo lithic settle ments on valley floors. Recent fieldwork in the region has expanded upon these earlier inves ti ga tions, result ing in the identi fi ca tion of a wider range of archae o log i cal sites (Acheulean through to Early His- toric peri ods) in a vari ety of depositional contexts (i.e., caves, rockshelters, open-air settings). Survey in the Jurreru River Valley revealed the presence of a volca nic tephra deposit, repre sent ing the Young Toba Tuff (YTT) of 74,000 years ago. Middle Paleolithic indus tries have been located in stratigraphic contexts above and below the tephra, thereby provid ing a unique oppor tu nity to exam ine stone tool technol ogy and human responses to this volca nic super-eruption. Although distinct “Upper Paleolithic” blade tool indus tries have yet to be reaf firmed through our cave inves ti ga tions, stratigraphic ob- serva tions indi cate that Late Pleisto cene stone tool assem blages (flake, blade, bladelet) and faunas are indeed present. Sur- vey results in dicate that many hundreds of unre corded rockshelters are present in the Kurnool District, a signif i cant pro- portion bearing rock art and surface microlithic in dustries. Rockshelter test exca va tions have indi cated buried assem - blages, with one rockshelter contain ing dated microlithic assem blages at least 34,000 cali brated radio car bon years old. The valley floors contain Neolithic, Iron Age and Early Historic period settle ments as well as Megalithic burial structures. Con tem po ra ne ous mid- to late-Ho lo cene cave and rockshelter set tle ments are in ferred here as rep re sent ing the ac tiv i ties of 120 M. Petraglia et al.

for ag ing pop u la tions. The ex traor di nary wealth of ar chae o log i cal re sources in the Kurnool Dis trict in di cates that the region can contrib ute substan tive infor ma tion about long- and short-term changes in human adap ta tions and social complex ity.

IN TRO DUC TION posits, and conduct survey in order to locate new The Kurnool District is one of India’s most sites. Based on the iden tifi ca tion of some sig nifi - im por tant ar chae o log i cal re gions, with nu mer ous cant ar chae o log ical sites in the Jurreru River Val - sites identi fied in the area over the past 150 years. ley and its vicin ity, a second season of field re - The region is partic u larly known for its Paleo- search was con cen trated in the valley. A com- lithic re cord. Co lo nial ex plor ers be gan re search in prehensive list of ar chae o log ical sites that were the dis trict in the mid-nine teenth cen tury, in ves ti - re-vis ited or iden tified dur ing the two first field gating the famous cave com plex at Billasurgam sea sons has been com piled (Fig. 1, Tab. 1). In this (Newbold, 1844). The caves were subject to arti cle, we will begin by exam ining the regional large-scale ar chae o log i cal ex ca va tions, in some ge ol ogy and geo mor phol ogy that pro vides the contexts reaching depths of 10 meters (Foote, context for the Pleisto cene and Holo cene sites dis- 1884a). These early explor ers observed that a covered in the region. Subse quently, we will ex- “Magdalenian” bone indus try was present, and am ine the ar chae o log i cal re cord of the area we that changes of envi ron ment were evi dent, as indi - have surveyed, looking first at cave sites in the cated by fauna as so ciated with for ests, grass lands western part of the Cuddapah Basin, then at the and swamps (Foote, 1884b; Lydekker, 1886). Jurreru River Valley open-air and rockshelter Later ex plor ers ob served that Paleolithic sites of sites, and fi nally at rock art sites. We will con - every period were repre sented in Kurnool (Cam- clude with a broad overview of the archae ology of miade, 1927). the region from the Pleisto cene through the Holo - The most com pre hen sive ar chae o log i cal cene, pre sent ing some hy poth e ses for fur ther work in the Kurnool District was conducted in the testing. 1960s and 1970s when system atic survey and ex- cava tion ef forts be gan in ear nest (e.g., Isaac, 1960; Thimma Reddy, 1968; Murty, 1974; Murty REGIONAL GEOL OGY and Thimma Reddy, 1975; Thimma Reddy, The study area falls within a dis tinc tive geo - 1977). Among the major findings of this re search logi cal terrain known as the Cuddapah Super- was the identi fica tion of blade indus tries in caves group, a crescent shaped basin of Protero zoic age, and open-air settings (Murty, 1979). Such find- surrounded by the Penin su lar Gneissic com plex ings were thought to fill the tem poral and indus - (Gupta et al., 2003). The Supergroup consists of a trial gap between the better known as sem blages of 6,000 meter thick se quence of al ter nating sand - the Mid dle Paleolithic and the Mesolithic (Murty, stones, shales and limestones, with evi dence for 1979). Along side Paleolithic in ves ti ga tions in the vol canic ac tiv ity towards the base of the succes - re gion, ar chae o log i cal re search was car ried out of sion, and the intru sion of ba sic sills (Lakshmi and Neolithic and Iron Age sites, result ing in the re- Babu, 2002). The east ern part of the basin shows cording of many settle ments (e.g., Sarma, 1967, the greatest degree of metamor phism and defor - 1968; Murty, 1989). A dis tinct Neolithic ce ramic mation (Saha, 2002). Be cause of the amount of ware type, known as Pathapadu (or Patpad) ware, meta morphism, shales have been turned into was defined from the site of Pathapadu, based on slates and sandstones into quartzites, making exten sive surface collections (Foote, 1916; them prized building mate ri als that are widely Allchin, 1962; Sarma, 1968). quarried today in the region. This ar ticle de scribes our own recent ar chae o - Our field project focussed on the rela tively logi cal research in the Kurnool District. The goal undeformed western part of the Cuddapah Basin, of our first season was to re-identify cave sites, consist ing of the Kurnool Group, a 500 m thick de ter mine the ex tent of their ar chae o log i cal de - suc ces sion of al ter nat ing quartzites, shales and Investigations in the Kurnool Dis trict, South India 121

Fig. 1. Lo ca tion of sites in the Kurnool District (see Ta ble 1). It should be noted that many ar eas re main unsurveyed, hence many other sites are present in this region limestones. The landscape is charac ter ised by a down the valley sides under the influ ence of grav- gently dipping plateau formed by a resis tant ity, and many of these tumbled blocks form caprock, the Panniam Quartz ite, which over lies a rockshelters with exten sive overhangs that have sequence compris ing the Owk Shale, the Narji been suit able for hu man hab i ta tion. Limestone, the Banganapalli Sandstone and the The limestone caves were mean while formed Tadipatri Shale, thought to repre sent transi tions by disso lu tion via subsurface and surface water between allu vial and intertidal depositional envi - flow, the cavi ties often forming along joints and ron ments (Lakshminarayana et al., 1999). Despite bed surfaces (Fig. 4). Variabil ity in water flow the rela tively undeformed nature of the sequence, and dis so lu tion pro cesses pro duced sig nif i cant conju gate fractures are widely devel oped (Fig. 2), differ ences in the height and width of caves and and faulting, al beit with limited dis place ment, has the number of connected passages. Caves forming cre ated a more complex structural pat tern in some in joints tend to be tall and narrow in shape, parts of the ba sin (Chatterjee and Bhattacharji, whereas those form ing along bed surfaces tend to 2001). Dipslope channels have cut some major be wide along their floors. The caves show the valleys through the Panniam Quartzite, thereby for ma tion of sta lag mites, stalagtites, brec cias, and forming major passes through the quartzite pla - sed i ments. Con tem po rary flu vial pro cesses are in- teau. Drainage has followed some of the conju - suffi cient to account for cave and valley forma - gate frac tures, forming steep-sided valleys cut tion, and the contem po rary drainage pattern ap- into the quartzite (Fig. 3). Wherever dipslope or pears to be underfit for the most part. The frac ture-con trolled val leys are cut into the quartz - pres ence of ar chae o log i cal ma te rial in caves, ite, very large quartzite boulders have moved rockshelters and on hillslopes indi cates that the 122 M. Petraglia et al.

Ta ble 1 Com pre hen sive cat a logue of Lo cal i ties Iden ti fied in the Kurnool Dis trict (2003–2004)

Map Local ity GPS/Elev.(m) Notes No. Cave com plex with four ex ca vated lo cal i ties (Char nel 1 Billasurgum N15°26.153'; E78°11.122'; Elev. 377 House, Purga tory, Cathe dral, Chapter House North). Rockshelter with unique beak-like overhang form, abun- 2 Bugganapalle I N15°27.448'; E78°11.066'; Elev. 448 dant microliths, rockart 3 Deyyapu Gundu I N15°20.398'; E78°08.410; Elev. 345 Rockshelter, some lithics, potsherds, rockart 4 Deyyapu Gundu, Ox N15°20.655'; E78°08.514'; Elev. 273 Rockshelter, some potsherds, wall construc tion, rockart Deyyapu Gundu, 5 N15°20.608'; E78°08.479; Elev. 311 Rockshelter, rockart Cross Deyyapu Gundu Hab ita tion site with abun dant ce ram ics and lithics in 6 N15°20.595'; E78°08.585'; Elev. 264 Hab i ta tion plowed field, in tact mega liths on hillslopes Rockshelter in re mote val ley, central view of val ley floor, 7 Dunnapotula Vanka I N15°27.946'; E78°12.174; 'Elev. 361 rockart 8 Genarden's Shel ter N15° 24.788'; E78°09.169'; Elev. 393 Rockshelter, rockart 9 Gundipalle I N15°32.368'; E78°19.028'; Elev. 264 Rockshelter, rockart 10 Gundipalle II N15°32.367'; E78°19.017'; Elev. 262 Rockshelter, rockart 11 Gundipalle III N15°32.529'; E78°18.889'; Elev. 232 Rockshelter, rockart Large cave with stone wall con struc tion (likely for trap - 12 Jinu's Ladder Cave N15°24.826'; E78°09.195'; Elev. 384 ping porcu pines); tall wooden and thatched ladder in cave chimney for uncer tain ethnographic function. Trenches 3 and 3A placed to obtain compos ite section below and above ash. This includes thick tephra depos its 13 Jwalapurum 3, 3A N15°19.338'; E78°08.029'; Elev. 253 and sig nif i cant ar chae o log i cal de pos its; see also Jwalapuram Local i ties 17 and 21. Central rockshelter along northern slopes, dense concen - tration of lithics (mostly microlithic), pot tery, groundstone 14 Jwalapuram 9 N15°19.864'; E78°07.788'; Elev. 266 tools on surface; 3.4m long profile with ceramic and pre-ceramic depos its, burial pits, rockart Western most rockshelter along northern slope, mostly de- 15 Jwalapuram 11 N15°19.939'; E78°07.628'; Elev. 277 stroyed by quarry ing, burials, lithics and potsherds, rockart Rockshelter with moder ate sur face scat ter of lithics 16 Jwalapuram 12 N15°19.927'; E78°07.672'; Elev. 281 (mostly micro liths) and pot sherds, ar chae o log i cal de posits, human skele tal parts, rockart 17 Jwalapuram 13 N15°19.888'; E78°07.753'; Elev. 281 Rockshelter with a few stone tools on sur face, rockart 18 Jwalapuram 14 N15°19.827'; E78°07.738; 'Elev. 281 Megalithic on surfaces in front of rockshelters Eastern most rockshelter along northern slope, a few 19 Jwalapuram 16 N15°19.790'; E78°07.950'; Elev. 306 lithics and potsherds on surface, rockart Tephra, mined. Stone tools on surface; tests indi cated ar- 20 Jwalapuram 17 N15°19.337'; E78°07.503'; Elev. 256 ti facts above tephra 21 Jwalapuram 19 N15°20.156'; E78°08.182'; Elev. 409 Rockshelter under quartzite escarp ment, one pictograph Dense concen tra tion of lithics on surface, below limestone 22 Jwalapuram 20 N15°18.918'; E78°07.835'; Elev. 258 raw ma te rial source; ar chae o log i cal de pos its with pre - pared cores, unifacially retouched tools Mined out tephra sec tion with lithics on surface; test with 23 Jwalapuram 21 N15°19.398'; E78°07.957'; Elev. 253 ar chae o log i cal de pos its above ash; see also Jwalapuram 3 and 3A. Investigations in the Kurnool Dis trict, South India 123

Ta ble 1 con tin ued

Map Local ity GPS/Elev.(m) Notes No. Mega lith, mul tiple col lapsed cists in a cairn, con sist ing of 24 Jwalapuram 24 N15°19.847'E78°07.711'Elev. 255 quartzite cobbles and boulders. Cists made of limestone slabs Destroyed megalithic structures due to modern soil quar - 25 Jwalapuram 25 N15°19.873'; E78°07.705'; Elev. 257 ry ing, pot sherds Large cut ting tools (Acheulean) ce mented in cal care ous 26 Jwalapurum Tank N15°18.637'; E78°08.231'; Elev. 259 de posit. Lithics along hillslopes and in ag ricul tural fields Jwalapuram Tank 27 N15°18.644'; E78°08.327'; Elev. 410 A few large cut ting tools in ag ricul tural field East Rockshelter with moder ate scatter of microliths, abundant 28 Katavani Kunta 1 N15°21.639'; E78°07.609'; Elev. 435 rockart 29 Katavani Kunta 2 N15°21.642'; E78°07.600'; Elev. 435 Rockshelter, no rockart 30 Katavani Kunta 3 N15°21.655'; E78°07.595'; Elev. 435 Rockshelter, rockart with elab o rate panel 31 Katavani Kunta 4 N15°21.763'; E78°07.478'; Elev. 472 Rockshelter, potsherds, rockart includes pot depic tions 32 Katavani Kunta 5 N15°21.770'; E78°07.429'; Elev. 456 Rockshelter, rockart (one depic tion) 33 Katavani Kunta 6 N15°21.818'; E78°07.176; Elev. 453 Rockshelter, rockart 34 Katavani Kunta 7 N15°21.798'; E78°07.184'; Elev. 474 Rockshelter, rockart, modern Hindu shrine 35 Katavani Kunta 8 N15°1.796'; E78°07.493'; Elev. 470 Rockshelter, some microliths on surface, rockart 36 Katavanii Kunta 9 N15°21.659'; E78°07.775'; Elev. 444 Rockshelter, rock art 37 Katavani Kunta 10 N15°21.636'; E78°07.803'; Elev. 445 Rockshelter, rockart 38 Katavani Kunta 11 N15°21.474'; E78°07.834'; Elev. 515 Rockshelter, rockart 39 Katavani Kunta 12 N15°21.866'; E78°07.262'; Elev. 464 Rockshelter, rockart 40 Katavani Kunta 13 N15°21.743'; E78°07.706'; Elev. 444 Rockshelter, rockart 41 Katavani Kunta 14 N15°21.749'; E78°07.709; Elev. 444 Boulder with Hindu linga engrav ing 42 Katavani Kunta 15 N15°21.530'; E78°07.876'; Elev. 464 Rockshelter, rockart (rem nant) 43 Katavani Kunta 16 N15o 21.534' E78°07.909'; Elev. 468 Rockshelter, rockart Katavani Kunta, Rockshelter, reli gious rockart, hundreds of Vishnu motifs, 44 N15°21.094'; E78°08.158; Elev. 393 Pilgrim 1 shrine with en graved feet slab Katavani Kunta, 45 N15°21.094'; E78°08.158'; Elev. 393 Rockshelter, re li gious rockart Pilgrim 2 Katavani Kunta, 46 N15°21.272'; E78°07.999'; Elev. 403 Rockshelter, re li gious rockart Pilgrim 3 Katavani Kunta, 47 N15°21.282'; E78°07.971'; Elev. 401 Rockshelter, re li gious rockart Pilgrim 4 Katavani Kunta, 48 N15°21.306'; E78°07.941'; Elev. 422 Rockshelter, re li gious rockart Pilgrim 5 Katavani Kunta, 49 N15°21.322'; E78°07.933'; Elev. 422 Rockshelter, re li gious rockart Pilgrim 6 Katavani Kunta, 50 N15°21.596'; E78°07.663'; Elev. 424 Rockshelter, re li gious rockart Pilgrim 7 Katavani Kunta, 51 N15°21.784'; E78°07.500'; Elev. 452 Rockshelter, re li gious rockart Pilgrim 8 Katavani Kunta, Quartzite bedrock with engrav ing of Nandi's (sacred ox) 52 N15°21.893'; E78°07.389'; Elev. 439 Pilgrim 9 head 124 M. Petraglia et al.

Ta ble 1 con tin ued

Map Local ity GPS/Elev.(m) Notes No. Katavani Kunta, 53 N15°21.312'; E78°07.923'; Elev. 400 Rockshelter, re li gious rockart Pilgrim 10 Kolamagalu 54 N15°32.301'; E78°19.255'; Elev. 258 Rockshelter, microliths, rockart Rockshelter Exca vated cave; Reddy 1976 reported stone tools, worked 55 Kottala Polmera Gavi N15°25.605'; E78°12.063'; Elev. 440 bone tools. Cave is a joint in limestone, very restricted in width, small terrace in front of cave Mega lithic dol mens along limestone hillcrest. Dol mens 56 Mogasarayanigondi I N15°19.934'; E78°08.464'; Elev. 262 suffer ing looting. Spoil piles contain Iron Age pottery, skel e tal re mains, burial goods Muchchatla Exca vated cave; Murty 1979 reported stone tools, worked 57 N15°24.879'; E78°07.235'; Elev. 398 Chintamanu Gavi I bone tools. Re-exam ined in 2003 Muchchatla Cave with multi ple pas sages. Near MCG I, re ported as a 58 N15°24.616'; E78°06.698'; Elev. 389 Chintamanu Gavi II Mesolithic site by Murty Muchatla 59 N15°24.578'; E78°07.369'; Elev. 470 Rockshelter, rockart on limestone Chintamanu Gavi IV Clas sic Neo lithic hab i ta tion with Pathapadu ce ram ics (Foote 1916; Allchin 1962; Sarma 1968). Potsherds in 60 Pathapadu 1 N15°19.812'; E78°09.996'; Elev. 239 plowed fields, no subsurface depos its, site appar ently destroyed Sanyasula Gavi, 61 N15°28.026'; E78°12.608'; Elev. 339 Cave with re cent hab i ta tion; ar chae o log i cal de pos its Cave 1 Sanyasula Gavi, Cave with re cent hab i ta tion; ar chae o log i cal de pos its, dis - 62 N15°28.025'; E78°12.621'; Elev. 339 Cave 2 tinct micromammal bone bed, ash bed 63 Yaganti 1 N15°1.197'; E78°08.494'; Elev. 366 Rockshelter, potsherds (includ ing Patpad Ware), rockart 64 Yaganti 2 N15°21.207'; E78°08.407'; Elev. 438 Rockshelter and cave, wall constructed in front, rockart 65 Yaganti 3 N15°21.219'; E78°08.364'; Elev. 517 Rockshelter, limestone slab mega lith cist burial, rockart Rockshelter, reli gious Hindu shrine with snake sculptures, 66 Yaganti 4 N15°21.176'; E78°08.385'; Elev. 366 rockart 67 Yaganti 5 N15°20.915'; E78°08.394'; Elev. 320 Rockshelter, rockart 68 Yaganti 9 N15°21.464'; E78°07.762'; Elev. 426 Quartzite boulder with Hindu linga engrav ing

valleys reached their general morphol ogy prior to the south end of the valley. Drainage is strongly hominin oc cu pa tion. controlled by structure, with a major sub-catch- The re gion is criss-crossed by a num ber of ment cutting through into the main valley at the river val leys, includ ing the Jurreru River Val ley. Jwalapuram Tank. A strongly indurated carbon ate The Jurreru River Valley is a major valley running duricrust at the Jwalapuram Tank ce ments coarse in an easterly course through the Kurnool Group, fluvial gravels. North of the main valley, a frac- with steep sides but a rel a tively wide bot tom (Fig. ture-controlled valley runs northwest from the 5). At the vil lage of Jwalapuram, a dolerite sill is Yaganti Tem ple. This is a steep-sided, V-shaped intruded into the shales on the southern side of the valley for the southeast ern half, but opens out into valley, forming a signif i cant ridge. A fault can be a wider-bottomed valley with cul tiva tion in its inferred be neath the allu vial cover of the val ley north west ern half. Tum bled quartz ite boul ders floor as the geolog i cal succes sion dips to the east form rockshelters on the valley sides. to the north of the valley and to the southeast in Investigations in the Kurnool Dis trict, South India 125

Fig. 2. Conju gate fractures in the Panniam Quartz ite. The frac tures through the quartz ite con trol the flow of drain age. In ad di tion, re treat of the es carp ment in each frac ture re sults in the pro duc tion of many boul ders. It is un - der these fallen boul ders that hu man hab i ta tion oc curred

Fig. 3. Ex ample of drain age in the con ju gate fractures of the Panniam Quartz ite. Steep-sided val leys form in the frac tures. Boul ders which have fallen from the quartz ite bed rock are vis i ble along the slopes 126 M. Petraglia et al.

Fig. 4. Lime stone caves formed by disso lu tion, Billasurgam cave com plex. This is the main en trance to the caves

Fig. 5. The Jurreru River Val ley showing steep val ley sides and a rel a tively wide bot tom Investigations in the Kurnool Dis trict, South India 127

CAVE RE CORDS ca vated ma te ri als were sub se quently lost, mak ing It has been esti mated that the thick limestone re-anal y sis im pos si ble (Cammiade, 1927: 7). units of the region may contain more than 100 Given the ab sence of clear ar chae o log i cal de pos - caves (Prasad, 1996), although most have not its in the caves, it was subse quently argued that been re corded and ex plored ar chae o log i cally. Ac- the bone objects may have been the result of the cordingly, we have made an effort to sys tem ati - ac tiv i ties of “dark ness-lov ing an i mals” such as cally record some of the best-known caves and to porcupines (Cammiade, 1927: 9), which were identify new sites. Our regional recon nais sance noted as abundant in the cave system (Foote, survey indi cated that cave morphologies varied 1916: 119). sig nif i cantly, likely in flu enc ing the de gree to The nine teenth cen tury ex ca va tions were re - which these envi ron ments were oc cupied by hu - portedly dug to depths of up to 10 m. Our inspec - man popu la tions. The large size and favor able tion of cave areas dug out by the Footes indi cate ecolog i cal posi tion of some of these contexts sug- that the major ity of the depos its were removed gest that they have po ten tial to yield cul tural de - within Charnel House and Purga tory Caves, al - posits. Jwalapuram Local ity 18, for exam ple, is a though some remnant de posits may be present on their terraces, under spoil heaps, and in the deeper very large cave complex with sev eral large en - 1 trances sit u ated just above a wide val ley floor. re cesses of the caves . As its name implies, the Though the cave has sedi men tary depos its (as in- Cathe dral Cave is an impres sive cave, massive in dicated by vil lager soil quarry ing) and several sur- size, hav ing a number of arch ways from a fallen face potsherds, little else is known about this lo- roof, with numer ous sta lac tites along its ceil ing. cal ity. In other cases, cave morphologies indi cate Our vi sual in spec tion of Ca the dral Cave in di cates that hu man ac tiv ity is prob a bly limited. For ex am- that the upper depos its were removed in the nine - ple, the Sanyasula Riv er side cave is a tall and nar - teenth century. However, it appears that substan - row crevice formed in the joint surface of a tial de posits remain, based upon inspec tion of an quartzitic unit, and the Gollagutta Gavi has a very open exca va tion trench. The trench appears to be re stricted height, as it is formed along a limestone the result of paleontological exca va tions con- bed seam. Though such spatially re strictive conducted in Cathe dral Cave in the 1960s, which re- texts may not be condu cive to human occu pa tion, sulted in the re covery of mamma lian, reptil ian, prior ex cava tions at the Kottala Polimera Gavi, and avian fossils (Prasad, 1996: 30). The open which is a nar row crev ice in limestone, indi cated trench is aproximately 6 m in depth and its walls that fau nal re mains and other paleoecological ev i- are well preserved on ac count of the heavy cal cifi - dence were present (Thimma Reddy, 1977). cation of the sedi ments. Inspec tion of the trench sug gests good po ten tial for re cov er ing anthropo- genic depos its, fauna and paleoenvironmental in - for ma tion. Al though lit tle in for ma tion is cur rently available about the forma tion of the Billasurgam The Billasurgam Cave Complex depos its, the brecciated sedi ments have produced The Billasurgam cave complex is a very large a rich range of fauna (Amphibia, Artiodactyla, and im pressive cave sys tem with a long history of Carnivora, Chiroptera, Crocodilia, Insectivora, ar chae o log i cal in ves ti ga tion. The com plex was Lagomorpha, Perissodactyla, Pholidota, Primates, first exca vated by Robert Bruce Foote, and his Reptila, Rodentia), in di cat ing wet en vi ron ments son, Henry Bruce Foote (Foote, 1884a, 1884b, with forests and grasslands (Lydekker, 1886; 1885). Exca va tions were conducted in three dif- Prasad and Yadagiri, 1986). The presence of fer ent caves within a sin gle large sys tem: Char nel Theropithecus in these caves has been in ter preted House Cave, Purga tory Cave and Ca thedral Cave as ev idence for a Late Pleis to cene ex pan sion of (Fig. 6). These early ex plor ers re ported a “Mag da - this Af ri can spe cies (Rus sell Ciohon, Terry Harri - lenian” bone indus try in the caves, and pieces son, pers. comm.), of poten tial im portance to our were classed un der a vari ety of cat ego ries, in clud - under stand ing of the dispersal of modern humans. ing awls, arrow heads, spear and harpoon heads, Ra dio car bon sam ples re port edly dated some of chis els, and scrap ers. Un for tu nately, all of the ex - the faunal remains to 50 kya (Prasad, 1996). 128 M. Petraglia et al.

Fig. 6. The Ca the dral Cave in the Billasurgam Cave Complex. Note the large size of the cave based on the in di - vidu als standing in the cave. Stalagtites and stalag mites are impres sive features. The paleontological exca va tions oc curred in the deep re cess on the up per right side of this cave

Chapter House North Cave an age was not assigned to Layer 1, the depos its Re newed ar chae o log i cal re search was un der - were clearly of Late Ho lo cene age as four pot - taken in the Billasurgam Caves in the 1970s by K. sherds were recov ered. The re cov ery of a “blade Thimma Reddy (1977), who conducted an exca - indus try” (160 of 951 lithic arti facts) in Layer 2 vation in an unex plored cave opening onto the (35–155 cm) was inter preted as evi dence of an northern edge of an area previ ously desig nated as “Up per Paleolithic”’ oc cu pa tion. the Chapter House (Lydekker 1886; Fig. 7). The stone tools were reported to be asso ci ated Thimma Reddy (1977) does not provide a name with a “bone tool indus try”, presum ably vindi cat - for this site and it is therefore des ignated Chapter ing Foote’s asser tion that humans were pres ent in House North Cave for the purposes this report. the caves. The stone tools, identi fied as a ‘crude’ The exca va tions resulted in the first published dis- blade indus try, were thought to repre sent occu pa - covery of stone tools in asso ci a tion with faunal re- tions that were in an early Up per Paleolithic con - mains at Billasurgam. The ex cava tions at Chap ter text. The bone tool indus try was collected from House North Cave produced a total of 1356 lithic two sep arate cave con texts (i.e., the Chap ter arti facts (Thimma Reddy, 1977). Most of the as - House North Cave and the nearby Kottala semblage consisted of flakes and chips (n = 981, Polimera Gavi). A total of 147 bone objects were or 73%) al though 240 ar ti facts were identi fied as classed as finished forms, worked bones, and tools (e.g., burins, points, borers, scrapers). While bone blanks, and 47 of the finished pieces were Investigations in the Kurnool Dis trict, South India 129

Fig. 7. View of Chap ter House North cave en trance. Thimma Reddy ex ca vated in this cave at Billasurgam in the 1970s and iden ti fied an “Up per Paleolithic” in dustry, re port edly con sisting of lithic ar ti facts and bone tools classi fied as points, chisel-edge tools, perfo ra tors, Reddy’s ex ca va tions). The silty sed i ments of spatulae, and scrap ers. While it is con ceiv able that Chapter House North Cave differ in compar i son some of the bones were the product of human ac - to the cal ci fied de pos its of Ca the dral Cave, in di - tivi ties, it has been argued, based on review of ar- cat ing differ ent depositional his to ries in the same ti fact drawings, that some of the Kurnool bone cave complex. Two explor atory 1x1 m units were tools were the result of ani mal contri bu tions, placed alongside Thimma Reddy’s origi nal echo ing Cammiade’s ear lier doubts (Petraglia, trench, though we did not reach the lower strata. 1995: 460). With respect to an inter preta tion for a Ob ser va tion of lithic items re cov ered from nat u ral ac cu mu la tion, it is in struc tive to note that the two units indi cated that it was sometimes dif fi - that the major ity of the re cov ered fau nal re mains cult to dis crim i nate nat u ral pieces from gen u ine (n = 910 of 949 items) are from Kottala Polimera cul tural ar ti facts. Sharp-edged lime stone spalls Gavi, a cave that is very re stricted in size and a sometimes shared attrib utes of cul tural pieces, as context that appar ently did not produce stone arti - observed in other cave contexts (e.g., Nash, facts. 1993). The presence of natu ral spalls led to the Upon our visit to Chapter House North Cave, pos si bil ity that some of the ar ti facts re cov ered the prior 2 x 3 m exca va tion trench was re-located from the pre vi ous ex ca va tions were nat u ral in or i- and the 3.5 m deep profile was observ able, and re - gin. Indeed, Thimma Reddy reported that some of corded as four main strata (Strata A–D, closely the ar tifacts were made of the same mate rial that approx i mat ing Layers 1, 2a, 2b, 3 in Thimma forms the cave walls and he also in di cated that a 130 M. Petraglia et al.

propor tion of the arti facts lacked clear signa tures These finds sug gest oc cu pa tion and gath er ing of human produc tion (i.e., bulb of percus sion, de- during the winter or dry season. Ziziphus fruit are lib er ate sec ond ary re touch). Ad di tion ally, the mainly available from Oc to ber to Feb ru ary/March stream bed running through the Billasurgam cave and were a common fruit throughout Neolithic com plex con tains abun dant ex am ples of nat u rally South India (Fuller et al., 2001a: 177–178). Al- flaked stone pieces identi cal to some of those though the pres ence of ceram ics implies that these found in the depos its. Never the less, a number of are Neo lithic in age, it is pos si ble that these were the re cov ered ob jects dis played flak ing char ac ter - received in trade by forag ing popu la tions, or else is tics that in di cate cul tural man u fac ture. A to tal of sea sonal vis i tors from farm ing com mu ni ties. 115 lithic ar tifacts was recov ered from the two Ce ram ics sug gest oc cu pa tion through the units, the major ity con sisting of flakes and chips. Second Millen nium BC. In total 33 sherds were Only six lami nar ob jects were re cov ered, two recov ered. These were studied alongside those bladelets (<2 cm) and four blades in Strata B and from Sanyasula Gavi, Jwalapuram rockshelters 9 C. Other tool types described by Thimma Reddy and 11 (see below), and two repre sen ta tive Neo- (e.g., burins, points, borers) were not recov ered, lithic sites to the Southeast, Rupanagudi (in north - al though the ex tent of the test units was limited. ern Cuddapah District, near the Kunderu) and The small faunal assem blage (n = 272) is do- Bilajapalle (in Jammalamadufu Taluk, south of minated by microvertebrate re mains, mostly ro - the Pennar) (for an intro duc tion to these sites, see dents, which be come rela tively more fre quent in Venkatasubbaiah et al., 1992; Venkatasubbaiah, the up per lev els. The few iden tifi able macroverte- 1998; Fuller et al., 2001b). The ceram ics from brates are from do mes tic cattle, pan ther, and a Chapter House North Cave were a mixture of large felid. A cow in ci sor in di cates that domes tic thick-walled handmade sherds, mainly consis tent ani mals were avail able from the surround ing re - with a Neolithic to early Iron Age date, and bur- gion; this is the only speci men from a domes tic nished tan/brown surface finishes (Fig. 8). The an imal iden tified from the Billasurgam Cave only rec og niz able form was an everted rim, neck- Com plex. The carni vore remains might point to less jar of a form typi cal of the Southern Neo- non-hu man pred a tors us ing Chap ter House North lithic. Cave as a den, although there is no evi dence of Only a sin gle small sherd of the dis tinc tive gnaw ing, car ni vore young, or other in di ca tions of black-painted red pot tery of Patpad ware was re - den ning ac tiv ity. Un for tu nately, the as sem blage is covered at Chapter House North Cave. It has a at pres ent too small to un ravel its po ten tially com- distinc tive fabric with medium to coarse quartz plex taphonomic history in further detail. and fine to medium grog, which does not match The small mollusc assem blage (n = 79) con- the fab ric of Patpad ware ex amples from other tains only landsnails of the genera Cryptozona, stud ied sites. These com par i sons in di cate that this Cyclotopois, and Pterocyclus. There are no indi - painted ware was being pro duced at sev eral cen - ca tions that hu mans col lected the landsnails; mi- tres in the region of Kurnool and Cuddapah but nor tem po ral shifts in as sem blage com po si tion ap parently all east of the Erramalai hills. It should proba bly reflect changes in the suitabil ity of the be noted that Patpad ware sites and ceramic occu - cave as snail hab itat as well as the snail col lect- pations in the Kunderu valley to the east and the ing/con sum ing ac tiv i ties of non-hu man pred a tors. Pennar to the south have been dated mainly to Archaeobotanical evi dence from the upper - 1700–1400 cal. BC (Fuller et al. 2007). most de posits (Layer 1) suggest occu pa tion of the Charcoal ob tained from flo ta tion sam ples cave sea son ally by for ag ers, while the presence of produced the first chronometric ages for the ceram ics place these in late Neo lithic (mainly af - Chapter House North Cave. Three sam ples date ter 2000 BC). Flota tion samples produced only the upper Stratum A depos its between 1487 and small quanti ties of charcoal and no crop plant re- 1708 BP, ca. 250–600 cal. AD (Table 2). The ear- mains. Charred and uncharred fragments of fruit liest ceram ics as so ciate with a date of 3995 BP stones (endocarps) were recov ered, includ ing in (2550–2470 cal. BC), al though it may be that this par tic u lar those from Balanites aegptiaca, and a char coal, or the tree it was from, pre dates any of few from Indian jujube (Ziziphus mauretania). the ce ramics by ca. 500 years. The three dates Investigations in the Kurnool Dis trict, South India 131

Fig. 8. Rela tive frequen cies of ware classes at Billasurgam, Chapter House North Cave, totalled across all con - texts (to tal no. of sherds = 33)

Ta ble 2 Radio car bon ages from Stra tum A of Chapter House North Cave

Sam ple Unit Stra tum Ma te rial d13C Age (BP) OxA-17359 3 A, Level 2 Balanites sp. seed 20.74 1487 ± 27 OxA-17354 3 A, Level 2 Ziziphus sp. seed 20.56 1550 ± 28 OxA-17357 1 A, Level 3 Ziziphus sp. 22.16 1708 ± 28 OxA-17427 3 A, Level 4 char coal 25.58 3995 ± 32 from the First Millen nium AD suggest contin ued dern era, produc ing stone tools, faunal remains, occu pa tions of this site, although it is unclear and a “hearth” with a date co in cident with the late which artifactual finds are to be as so ciated with Upper Paleolithic. MCG was a promi nent site in these. Three sam ples from Stra tum B and C pro- Murty’s (1979) Upper Paleolithic synthe sis, duced unex pect edly young ages ranging from mo- which at tempted to dem on strate that a blade and dern to 570 BP, likely indi cat ing stratigraphic dis - bone indus try was present in India (Murty, 1979). turbances and downward migra tion of charcoal. Surveys of the region around Billasurgum re- Unfor tu nately, at this stage in our dat ing program vealed that nu mer ous cave and open-air sites were the inferred Upper Paleolithic age range of the present (Murty and Thimma Reddy, 1975) includ - lower-ly ing depos its could not be confirmed. ing a cave ad ja cent to MCG, which con tained abundant microlithic indus tries (Murty, 2003). MCG sits in the es carpment of a small and Muchchatla Chinamanu Gavi (MCG) winding limestone canyon, with three entrances M.L.K. Murty’s (1974) exca va tion at Much- sepa rated by a length of 5 m along the es carpment. chatla Chinamanu Gavi rep re sents a pi o neer ing The three en trances have pas sages that even tu ally work in Indian prehis toric studies as this was the join each other at a deeper inner chamber. The first cave to be exca vated in Kurnool in the mo- passages are narrow (ranging from 0.5 to 1.2 m 132 M. Petraglia et al.

wide) and even after exca va tion, the height of the gations were in agreement with the well-docu - cave from floor to roof was usually no greater mented strati graphic ob ser va tions and ar ti fact than 1.25 m. The thickness of the sedi ments in the findings de scribed in the MCG publi cations. cave ranged from a maxi mum of 40 to 85 cm. The The small ver tebrate as semblage (n = 72) size constraints of the passages and the dark, inte - con tained only three microvertebrate re mains; the rior posi tion of the inner cham ber were likely im - macrovertebrates include four-horned ante lope, por tant fac tors in con strain ing hu man ac tiv ity. hare, and porcu pine, and an nilgai-sized ungu late. Al though ex ten sive ex ca va tions were con- Landsnail remains are like wise rare. Our spe cies ducted on the terrace in front of the cave, in the list from MCG1 compares well with that re ported pas sages, and in side the in ner chamber, rela tively by Murty (1985) from MCG II. In contrast to few stone ar ti facts were re cov ered, to tal ling only other cave as semblages from the area (Sanyasula 223 pieces (Murty, 1974). The stone arti facts Gavi Caves 1-2, Chapter House North), microver- were accom panied by 2,003 “bone-tools” and 290 tebrate and mol lusc re mains are extremely rare faunal remains. Although a Neolithic bowl was and lim ited to the upper most level (A1). The lack found in side the cave, all other ar tifacts and mate - of microvertebrates, in partic u lar bats, is surpris - rial re mains were con sid ered to be Up per ing as the cave sur face was re ported to be cov ered Paleolithic in age. The Late Pleis to cene tempo ral with bat guano (Murty, 1974: 201; Prasad, 1996: esti mation was confirmed by a thermolumines- 32). cence (TL) date of 17,390 ± 10% BP from a “fire While MCG pro vides im por tant in for ma tion place” (Nambi and Murty, 1983). Murty’s identi - about human occu pa tion in the region, our re-ap- fica tion of a bone tool indus try has contin ued to praisal indi cates that more infor mation is needed play a role in dem onstrat ing the exis tence of to de ter mine the age of the caves and to char acter - worked bone in Upper Paleolithic contexts (e.g., ize the stone tool tech nology. While a Late Pleis - Sali, 1989; Raju and Venkatasubbaiah, 2002). to cene age of the de pos its is con ceiv able, the age How ever, as is the case in the Chap ter House of the assem blages can not be confi dently pin - North Cave, it is possi ble that the bone accu mu la - pointed as only one TL date is available. It is pos- tions resulted from a more com plex taphonomic sible that the blades de rived from “fluted” cores his tory, in clud ing con tri bu tions from an i mals may date to any time period within the Late Pleis - (Petraglia, 1995: 460). to cene or even the Ho lo cene. The re cov ery of a The goal of our ini tial work at MCG was to Neolithic bowl at MCG and the identi fi ca tion of place the site in landscape context by conduct ing a rock art and microlithic indus tries in other caves re con nais sance sur vey of the lime stone can yons. and rockshelters suggest that the oc cupa tion his- Pre lim i nary sur vey in di cated that MCG was not tory may be more com pli cated. An addi tional ca- an iso lated oc cur rence. Other caves (e.g., MCG veat is that while many of the il lus trated stone II-III, V) and rockshelters (MCG II, IV) were tools are ap par ently gen u ine ar ti facts, some level pres ent in the area. Ar tifacts were ob served on the of caution is warranted in distin guish ing cul tural sur faces of some cave floors and rock art was and natu ral pieces. Re view of illus trated pieces present along the wall of a rockshelter (MCG IV) and pre lim i nary ex am i na tion of ar ti facts on dis- and on the inside of one cave (MCG V). play at the Deccan College museum indi cate that We placed test ex cava tions in one pas sage of some pieces may be the re sult of natu ral spalling, MCG, in a pas sage de scribed by Murty as “de - as may also be the case in the Chap ter House posit left undis turbed”. The stra tigra phy was con- North Cave. Ev idence that many pieces may be sis tent with the three main layers de scribed in ear - natu ral in ori gin is suggested by a number of ob- lier exca va tions. The maxi mum depth of the ser va tions, i.e., most arti facts are made from the depos its was shallow, measur ing only 40 cm from limestone that forms the caves, the arti fact de - the sur face to the limestone floor. Ar tifacts were scriptions do not indi cate the presence of distinct recov ered in only one unit, and consisted of two tech no log i cal at trib utes (e.g., bulb of per cus sion), flakes and a blade, con sis tent with the rel atively flakes de tached from ‘cores’ do not show any ev i- low den sity of cul tural ma te ri als re cov ered in the dence of prep ara tion, and few ar tifacts show sec - ear lier ex ca va tions. The re sults of the trial in ves ti - ond ary retouch. Investigations in the Kurnool Dis trict, South India 133

The Sanyasula Gavi Com plex

The Sanyasula lo cali ties oc cur as a clus ter of caves situ ated in a sub-valley, removed from modern vil lages. The caves are fre quently noted in the regional liter a ture, though they have not been pre vi ously sub ject to ar chae o log i cal in ves ti gations. We recorded five caves, consist ing of three large caves (Sanyasula Caves 1–3), one cave chimney lead ing to an in ner recess (Sanyasula Chim- ney), and one formed within a joint surface (Sany- asula Riv er side Cave). The caves were sit u ated along a single south facing limestone cliff face, sep arated from each other by a dis tance of about 50 m. Upon vis ita tion, it was clear that the caves were being ac tively used by saddhus, or holy men, as indi cated by small Hindu shrines and paintings on the Sanyasula Chimney Cave. Sanyasula Caves 1 and 2 were in hab ited re cently as in di cated by hearths and shattered limestone debris on their surfaces (Fig. 9). Inter views with vil lage pasto ral - ists in di cated that saddhus oc ca sion ally visit and live in these remote caves. Ethnographic and ethnoarchaeological accounts in di cate that for ag - ers visited the caves in order to trap ani mals such Fig. 9. Sanyasula Cave 1, showing ev i dence of re - cent hab i ta tion. Note the hearth, with ash and un burned as porcu pines and collect honey from beehives wood and note the ar range ment of large limestone slabs (Murty, 1985). around the hearth Follow ing survey of the sub-valley, test exca - vations were conducted in Caves 1 and 2 in order to de ter mine if any ar chae o log i cal de pos its were present. A 1 × 1 m unit in Cave 1 reached a depth of 1.5 m, but large boul ders pre vented further Ceram ics from the upper units of both caves excavation. Arti facts in the top 50 cm included were simi lar in hav ing a mixture of Neo lithic and ceramics. A total of 116 lithic ar tifacts was recov - later types, although Neolithic types dom inated. ered in the pre-pot tery levels, and included blades From both units a total of 269 sherds was recov - and bladelets. A sandstone quern found in the up- ered and studied, of which a signif i cant number per lev els of Cave 1 in di cates the pos si bil ity for were of the dis tinc tive Patpad painted ware (Fig. plant process ing activ i ties. In Cave 2, a 1 × 2 m 10). A few unbur nished black-slipped, red-pol- unit was termi nated at 1.6 m depth on account of ished and coarse red ware sherds are likely to be com pact sedi ment that could not be pene trated in later from late Iron Age or his toric times, but most this lim ited ex ca va tion. The most prom i nent stra- of the pot tery fits within a Neo lithic as semblage, tigraphic fea ture in the 1 × 2 m unit was a bone such as the high number of tan/brown burnished bed, consist ing of large quanti ties of micromam- sherds (as at Chapter House North Cave). Forms mal fauna and an ash layer. As in Cave 1, pot- include typi cal Neolithic open bowls and a necked sherds were re covered in the upper most levels of jar of later Neo lithic type. Of par tic u lar note is Cave 2. Among the 45 lithic arti facts were blades that the Patpad sherds could be divided into two and bladelets. Radio car bon sam ples were ob- fabrics, both of which are known from the Neo- tained from Cave 2, provid ing esti mates of 1159 ± lithic site of Rupanagudi in the Kunderu river val - 30 BP for Stra tum A and 3515 ± 35 BP for Stra- ley. At Rupanagudi at least two other Patpad ware tum B (ca. 1900–1800 cal. BC) (Table 3). fab rics also oc cur. The in hab itants of these caves 134 M. Petraglia et al.

Ta ble 3 Radio car bon ages for Sanyasula Cave (from Fuller et al., 2007, Online Supple ment Table 4 [http://www.antiq uity.ac.uk/projgall/fuller/])

Sam ple Con text Level Ma te rial d13C Age (BP) R 28680/32 Sanyasula Gavi, near top A-2 Wood charcoal -29.1 1159 ±30 R 28680/33 Sanyasula Gavi, earli est pot tery B-5 Seed: Vigna sp. -22.58 3515 ±35 BA04397 Sanyasula Gavi, earli est pot tery B-5 Seed: Ziziphus cf. mauretania Not reported 3505 ±40 BA04398 Sanyasula Gavi, in tru sive frag ment B-6 Black wood, bark Not reported MOD ERN

Fig. 10. Rela tive frequen cies of ware classes at Sanysula Gavi, totalled across all con texts (total num ber of sherds = 269)

pre sum ably ob tained ce ram ics from the val ley, ric known in quantity from Sanganakallu Neo - and one or two consis tent fabrics suggest that they lithic sites in Bellary Dis trict (cf. Boivin et al., were in contact with only one or a few pot tery- 2005). This implies that there was some trade to pro duc ing vil lage com mu ni ties in the plains. By the west, al though Kurnool ceramic types pre- con trast the Neo lithic vil lage com mu ni ties, like dom i nate. Rupanagudi were obtain ing ce ram ics from a wide Of the many verte brate remains (n = 3836) net work of other com mu ni ties, rep re sented by a from the Sanyasula Caves, about 97% are from wider range of fabrics. This suggests that the dy - small-sized ro dents (pre dom i nant at Sanyasula namics of so cial networks were rather dif fer ent Gavi 2), bats (predom inant at Sanyasula Gavi 1), amongst the Neolithic plains com muni ties and be- birds, am phib i ans, and rep tiles. In ter est ingly, only tween the isolated Erramalai cave-dwell ers and wild taxa have been posi tively identi fied in the adja cent vil lages. Also of note is a single sherd small macrovertebrate as semblage – nilgai (Bose- with a silvery-grey burnished finish and a dis tinct laphus tragocamelus), sambar (Cervus uni- olive-grey paste with mica and grog temper, a fab- color), four-horned ante lope (Tetracerus qua- Investigations in the Kurnool Dis trict, South India 135 dricornis), a small bovid, a small fox (Vulpes sp.), Ta ble 4 black-naped hare (Lepus nigricollis), mongoose Archaeobotanical ev i dence (Herpestes edwardsi), bandi coot rat (Nesokia from Sanyasula Caves sp.), and porcu pine (Hystrix in dica). These sites produced a small landsnail assem blage (n = 303) Sanya. Sanya. Sanya. that includes a mix of the genera Cryptozona, 2.2.B-5 2.2.B-6 TR 3/A/2 Cyclotopois, and Pterocyclus. Flot. vol. (mL) 84 60 80 Using the rel ative frequency of micro- versus Flot. weight (g) 43 29.7 30 macro-verte brate remains as a rough index of hu- Vigna radiata (cot y le don) 1 man oc cu pa tion, these caves have com ple men tary Vigna sp. (frag ment) 1 sequences of human occu pa tion and use. The Sa- nyasula Gavi 1 was used more regu larly during Macrotyloma (fragment) 2 Setaria/Brachiaria/ level B at a time when Sanyasula Gavi 2 was pri - 1 mar ily a roost, while Sanyasula Gavi 2 was used Echinocloa (grain) more in ten sively by hu mans dur ing the ac cu mu la- Brachiaria ramose (grain) 3 cf. Hordeum vulgare tion of level A at a time when Sanyasula Gavi 1 1 was lit tle used. We suspect that both sites were oc - frag ments ca sion ally used by lo cal hunter-gath er ers who fruit/nut indet. fragments 1 might have exchanged wild products for domestic Ziziphus sp. fragments ++ 3 1 products from villages. Fruit pedicel 1 Flota tion was carried out on sedi ments from Indet small le gume 2 Sanyasula. Due to the long walking distance to the site and rugged terrain, only rela tively small vol- Asteraceae type 1 umes could be carried from the caves, and most cf. Convulvulaceae 1 samples yielded no identi fi able fruit re mains and Indet. fragments ++ 31 only very small char coal frag ments. Nev er the less Dung fragments 7 at least two crops were pos itively iden tified (Ta - ble 4), includ ing browntop mil let (Brachiaria ramose) and mungbean (Vigna radiata), two of THE JURRERU RIVER VAL LEY the predom inant staples of the South Indian Neo- Our team vis ited the Jurreru River Val ley lithic (Fuller et al., 2001a; 2004). A possi ble bar- based upon a geolog i cal publi ca tion that noted the ley frag ment was also pres ent, as was the In dian presence of volca nic tephra depos its in the valley ju jube fruit. A direct AMS date on the mungbean and ‘hominid’ remains in the Mogasarayanigondi (R 28680/33; 3515±35 BP) and another on ju jube cave (Rao and Rao, 1992)2. Inspec tion of the area (BA-04397; 3505±40 BP) confirmed the Neo- showed that substan tial quanti ties of ash were ac- lithic as soci a tion of these crops and put back the tively being mined by vil lagers for com mercial Neolithic occu pa tion in these caves to ca. 1900 purposes (Fig. 11). A recon nais sance around the BC. It should be noted that this repre sents the ear- tephra min ing op er a tions re vealed the pres ence of liest Neo lithic re mains and crop re mains in lithic ar ti facts as so ci ated with dif fer ent sed i men - Kurnool or Cuddapah districts, where the wide- tary depos its. Published sources, in fact, indi cated spread Neo lithic villages with dis tinc tive Patpad the high ar chae o log i cal po ten tial of the re gion, in- ware have been dated from 1700–1400 BC (Fuller cluding the pres ence of Paleolithic local ities et al., 2007). This suggests that the cave’s oc cu- (Cammiade, 1927; Isaac, 1960), the Pathapadu pants may have ac quired crops and some unde- Neolithic mound site (Allchin, 1962)3, Iron Age corated, less dis tinc tive pot tery initially from es - megaliths (Krishna Sastry, 2003: 125), and his- tab lished Neolithic farmers to the West of the toric period tem ples, includ ing the 15th century Erramalai hills, while the bulk of the Patpad ware Sri Yagatiswamy Temple at Yaganti. The follow - sug gests that con tin u ing in ter ac tion shifted to ining over view sum marizes some of the most prom - clude newly es tab lished Neo lithic set tle ments to inent sites identi fied during survey of the Jurreru the east af ter 1700 BC. River Val ley. 136 M. Petraglia et al.

Fig. 11. Vil lag ers mining the vol ca nic ash in the Jurreru River Val ley. Min ing trenches has ex posed the bur ied vol ca nic ash. In the back ground, the piles con sist of pure ash which has been screened of its de tri tus. In the fore - ground, the piles con sist of car bon ate and rhizoliths, which have been removed from the ash. This min ing ac tiv ity is leading to the destruc tion of the ash and asso ci ated archae o log i cal sites. Note also the quartzite escarp ment and boul ders in the back ground, which form the north ern slope of the val ley

The Jwalapuram Tank standing pools of water. The arti facts do not ap - pear to be transported by allu vial processes, Survey was performed along the southern hill though the tools are in shallow de posits that are flanks in the Jurreru River Val ley. The re connais - not condu cive for study of their origi nal spatial sance was con ducted at the break in slope where a pat terns. The LCTs are of dif fer ent sizes and river, the Kanigal Eru passed, and where a 1946 shapes, con ven tion ally clas si fi able as handaxes, tank was constructed. Large Cut ting Tools knives, and choppers (Fig. 13). The tools were (LCTs) of the Acheulean were found along the made on quartzite and limestone, raw mate ri als base of this north facing hillslope, distrib uted that are readily avail able in the vicin ity of the across an area measur ing 1 km east to west. The finds. Some of the tools are made on second ary ar ti facts were found in un con sol i dated ‘fanglome- clasts, such as quartzite cob bles, whereas oth ers rate’ de pos its and ce mented into con sol i dated cal - are manu fac tured on flakes. One handaxe is man - care ous depos its capped by a resid ual tephra (Fig. ufac tured on a large limestone slab, which is remi - 12). The arti facts appear to have been depos ited in niscent of the manu fac tur ing technique used in the col lu vial set tings and in set tings char ac ter ized by Isampur Quarry (Petraglia et al., 1999). Investigations in the Kurnool Dis trict, South India 137

Fig. 12. The Jwalapuram Tank de posit showing ce menta tion. The hammer marks the lo ca tion of a Large Cut ting Tool which is in situ

Fig. 13. Large Cut ting Tool recov ered from ce mented de posits at the Jwalapuram Tank. This ar ti fact was man u - fac tured from the nearby lime stone beds 138 M. Petraglia et al.

Fig. 14. View of an ash sec tion at Jwalapuram ex posed by modern quarry oper a tions. Test unit is be ing placed be - neath the ash

Jurreru River Val ley Tephra Depos its Lo cal ity 3 towards the northern quartzitic hill- The tephra depos its in the valley bottom are slopes indi cated the presence of a contin u ous scat- im pres sive sed i men tary for ma tions (Fig. 14). The ter of lithic ar tifacts. The aim of the ini tial ex ca va - tephra depos its were not observ able at ground sur- tions was to deter mine whether subsurface ar- face and could only be viewed in sec tion in mod - chae o log i cal de pos its were pres ent and to exa- ern ash quar ry ing op er a tions. Sur vey of ex posed mine their rela tion ship with the tephra. The fol- sections along the Jurreru River indi cated that the lowing descrip tions are the main findings of our ash could be found over a distance of approx i - ini tial in ves ti ga tions. mately 9 km (Fig. 15), with the exposed ash thickness ranging from 15 cm to 2.46 m. The ash was Jwalapuram Lo cal ity 3 thickest near the village of Jwalapuram, in the Lo cality 3 was the lo cation of the deep est ash central portion of the valley, and shallow along deposit in the val ley, measur ing 2.46 m in thick- the valley floor margins. ness (Figs 16, 17). Electron probe microanalysis Ar chae o log i cal field work con sisted of sys - (EPMA) of volca nic glass shards from the Jwala- tematic sur face col lec tion and test ing of strati - puram tephra indi cated that the ash is a distal de- graphic sec tions at four main lo cal ities in the cen - posit of the YTT of 74,000 years ago (Petraglia et tral portion of the valley (Jwalapuram 3, 17, 21, al., 2007). Op ti cally stim u lated lu mi nes cence Well Sec tion). Sys tem atic sur face col lec tion from (OSL) dat ing from ar chae o log i cal lay ers be low Investigations in the Kurnool Dis trict, South India 139

Fig. 15. Depth re cord ings for vol ca nic ash in Jurreru River Val ley. The depths were re corded from the vil lag ers mining trenches placed across the val ley. Note the loca tion of the thick est ash de posits in the south-cen tral area. The mod ern river cuts through the for mer ash de posits

(JLP3A-200) and above (JLP-380) the ash re- thick. It is overlain by a fining-up lam inated se- turned ages of 77 ± 6 kyr and 74 ± 7, respec tively quence of very fine-grained sedi ment that culmi - (Petraglia et al., 2007). nates in a dark hori zon form ing a prom inent and Several lines of evi dence, nota bly in the form lat erally exten sive hard band within the section. of sed i ment de for ma tion struc tures, sug gest that This is a pattern repeated an addi tional four times the tephra ini tially accu mu lated on a wet clay sub - higher up in the section. strate, prob a bly in a lac us trine set ting. A dis tinct A number of features, includ ing mudcracks, layer of ash, 4–5 cm thick, was identi fied at the and possi bly signs of micro bial mats, suggest that base of the tephra sequence, suggest ing that it rep - these rep re sent cy cles of des icca tion of the lac us - resented a primary fallout deposit. Above the trine en vi ron ment. Struc tures rem i nis cent of rip- basal ash layer is a homo ge neous layer of light ple marks were observed, though these could well grey ash up to 80 cm thick. This was presum ably be diagenetic features. Numer ous roots, both fos - still being depos ited within a water body, but the silised rhizoliths and unfossilized root casts, were deposit may have also formed by reaching the retrieved (Fig. 18). Im pressions of what could pooled water through aeolian or aque ous trans- possi bly be small and intri cate root networks were port. The ash is too thick to be primary fallout clearly seen on the surface of most of the hard hor- from the volca nic eruption plume, and hence it is izon tal lay ers. The upper se quence shows an consid ered, with much of the overly ing se quence, abrupt transi tion from grey ash to orange-brown to be the result of remobi li sation and redeposition. sed iment, the later appar ently an ash-rich silt There is no evi dence, at least on the basis of visual horizon. This appears to indi cate a signif i cant in spec tion, for in cor po ra tion of sig nif i cant change in depositional regime rather than pedo- amounts of any mate rial other than ash. The upper gene sis or diagenesis. The ash-rich nature of these boundary of this layer is a very distinct contact sed iments in di cates that there is un likely to be with an appar ently finer-grained, brown-grey, much of a time inter val between the two units. A pos si bly clay-bear ing layer, ap prox i mately 20 cm pos si ble in ter pre ta tion for this ash-rich ho ri zon is 140 M. Petraglia et al.

Fig. 16. Com pos ite strati graphic sec tion of Lo cal ity 3. This sec tion dem on strates the ex is tence of an ar chae o log i - cal ho ri zon be low the ash, brack eted by the 74 kyr old ash and an OSL sam ple dat ing to 77 kyr. Note the thick ness of the ash. Pri mary ashfall ex ists at the base of the ash sec tion. The bands may rep re sent redeposition events oc cur - ring soon af ter the ash fall. Oc cu pa tion oc curred soon af ter set tling of the ash on the val ley floor

a perma nent drying up of the lake, per haps asso ci - Un der lying the ash, we ex cavated a sec tion ated with cli matic change or the geomorphologi - measur ing 3.6 m in depth (Fig. 19). The depos its cal/hy dro log i cal im pacts of the ac cu mu lated te- were composed of three main strata (Fig. 16). phra. The up per most de pos its rep re sent flu vial ac- Noteable in this section was a pebble conglom er- tivi ties in the form of small pockets of sand and ate with ar chae o log i cal ma te rial, dated by OSL to peb bles. ~77 kyr BP. Investigations in the Kurnool Dis trict, South India 141

Fig. 17. View of ex ca vated pro file in ash, Jwalapuram Lo cal ity 3. Note se ries of hard hor i zon tal bands, which may repre sent a series of mon soonal events after the ash fall

Fig. 18. Ex amples of rhizoliths re cov ered from a trench in Lo cal ity 3. The rhizoliths are very abun dant in the ash pro vid ing in for ma tion about veg e ta tive com mu ni ties 142 M. Petraglia et al.

Fig. 19. Ex ca va tions be low the ash at Local ity 3. Ash is present along the side wall, above the trench. The lower ar chae o log i cal ho ri zon was be low this ash de posit

Jwalapuram Lo cal ity 21 the under ly ing sedi ments, dating to 77 ± 6 kyr BP. Local ity 21 is com posed of a 2.3 m deep The as semblage contained 215 ar tifacts as well as trench, with six main strata above the ash (Fig. a piece of red ochre which shows striations from 20). Stratum F is an orange silt deposit, and Strata use. This stone tool assem blage comprises a fac - B–D were rec ognized on the ba sis of vari ations in eted uni di rec tional core and nu mer ous flakes clay color and calcium carbon ate con tent. Thin made predom i nantly from limestone (60%), quar- bands of gravels are located within the clay indi - tzite (22%), and chert (11%). There is moder ate cat ing some fluvial activ ity, likely in the form of prepa ra tion of flake platforms through overhang sheetwash. Localised changes in depositional re- removal (43%) and facet ing (8%), suggest ing that gime were in di cated by silt in Stra tum A and these flakes were sometimes struck from pre pared sandy clay in Stratum E. OSL ages of ca. 38 kyr cores simi lar to those found at the site. Flakes are BP (JLP21B-80; JLP21B-30) were obtained from typi cally small at 34 ± 18mm and squat (mean L: this deposit. W = 1.3) with a single dor sal ridge. The pre-Toba assem blage contains a small num ber of blades and microblades up to 4 cm in length (2.8%) as well as Ar ti fact as sem blages as so ci ated with tephra a crested flake that may have been cre ated to set de pos its up ridges on a core. A sizeable propor tion of Lithic as semblages below the ash were chro - flakes (11%) were retouched into notches, infor - nolog i cally bracketed by the ~74 kyr BP YTT and mal scrap ers, a heavily re touched blade, and a Investigations in the Kurnool Dis trict, South India 143

Fig. 20. Stra tig ra phy of Jwalapuram Lo cal ity 21 burin spall. Re touched flakes are typi cally quite = 1.5), typ ically with a sin gle dor sal ridge, fre - heavily reduced (Kuhn’s (1990) GIUR = 0.8, % quent overhang removal (25%) and rare platform margins re touched = 80%). In addi tion to these facet ing (1.6%). Retouched items also make up a ar tifacts, a small num ber of re touched cor tical and sizeable portion of the assem blage (16%), includ - tab u lar pieces of limestone were re cov ered that ing burins and side- and end-scrapers, and not- bear strong resem blance to pieces re cov ered from ches. Over all re duc tion in ten sity is low for re - lo cal i ties else where in the val ley (i.e., Lo cal ity touched items (Kuhn’s GIUR = 0.05, % margins 20). This assem blage falls within the Indian Mid- retouched = 37%). A single blade and four micro- dle Paleolithic as iden tified else where (Misra, blades are also pres ent in the as sem blage (4.6%). 1989, James and Petraglia, 2005). A to tal of 131 ar tifacts were recov ered at Lo - The post-Toba layer at Lo cal ity 3, op tically cal ity 21 from a simi lar post-ash ma trix to that dated to 74 ± 7 kyr BP, contains an assem blage of seen at Local ity 3, although as noted the Local ity 108 stone arti facts that occur throughout the or- 21 sedi ments date to ~38 kyr BP. The Local ity 21 ange sandy stra tum. The assem blage comprises as sem blage dif fers in some im por tant re spects multiplatform cores, faceted sin gle plat form cores from Local ity 3 which is to be expected given and core fragments. Limestone makes up a more than 35,000 years between them. Chalce - smaller propor tion of the assem blage than below dony dom inates the assem blage (35%), fol lowed the ash (32%), around double the amount of chert by chert (29%) and limestone (27%). Com plete is found above the ash than below it (25%), chal- flakes are once again small (19 ± 9 mm) and squat cedony appears for the first time (23%), while (L: W = 1.3), with an aver age of one dorsal ridge, quartzite is around half as abundant as below the little cortex cov er age (7%) and little plat form ash (12%). Like the assem blage from below the prep a ra tion (6% fac et ing and 10% over hang re- ash, flakes are small (21 ± 16mm) and squat (L: W moval). A very high propor tion of flakes are re- 144 M. Petraglia et al.

Fig. 21. View of Jwalapuram Lo cal ity 20. Test ex ca va tions in south ern core are be ing placed in the plowed fields. Note the hillslopes to the north, which are com posed of quartz ite boul ders which served as rockshelters (e.g., Jwalapuram Lo cal ity 9)

touched into side- and end-scrapers (38%) of vari - slope, the surface arti facts were found in plowed ous kinds, as well as burins and notches. Retou- fields extend ing 100 m to the north. Surface arti - ched flakes overall show low levels of reduc tion facts in the western por tion of the site were sit u - (GIUR = 0.04, % margin retouched = 45%). The ated in an aca cia for est along col lu vial hillslope Local ity 21 assem blage contains two blades and sur faces and a ped iment. two microblades, suggest ing very low lev els of The large con cen tra tion of sur face ar ti facts produc tion for these kinds of arti facts. pres ent at Lo cality 20 was re lated to the pres ence of a plen tiful raw mate rial sup ply. A si liceous out - crop is present at the mid-slope of the hill, consist - Jwalapuram Lo cal ity 20 ing of a distinc tive dark greenish-gray limestone Jwalapuram Local ity 20 was identi fied during (Fig. 22). From the mid-slope to the base of the surface survey along the southern hill slopes of hill, a contin u ous scatter of angu lar and blocky the Jurreru River Valley (Fig. 21). The locale is clasts of the weath ered bed rock was pres ent. The approx i mately 1 km west of the break in slope major ity of the ar tifacts ap peared to have been pres ent at the Jwalapuram Tank. An aeri ally ex- produced from the second ary clasts rather than di- tensive scatter of arti facts was found along the rectly from the limestone outcrop. base of the collu vial hillslopes, measur ing a dis- A pedestrian sur face col lection (Transect 1) tance of over 200 m east to west. From the base of was performed to re trieve a sample of the sur face Investigations in the Kurnool Dis trict, South India 145

Fig. 22. Source of si li ceous lime stone bed rock on the south ern hillslope at Lo cal ity 20. It is from this bed rock de - posit that weath ered tab ular pieces form. Tab ular and an gular pieces commonly served as cores in the Middle Paleolithic arti facts. Transect 1 was about 10 m in width and com mon plat form type was plain (n = 32; 71%), placed in an east to west direc tion, 50 m in the followed by faceted forms (n = 12; 27%) and a plowed fields and 50 m along the forested hill- single di hedral plat form. Of the formed tools re- slope. A series of 1 × 1 m units were then placed covered during surface walk over, the great major - in a cross-pattern across the site (i.e., north-south, ity are unifacially re touched, with all the unifaces east-west) to ex am ine vari ations in land scape fea - made on flakes with the excep tion of a single tures and stra tig ra phy. piece made on a round pebble. A total of 272 arti facts was col lected from Transect 1, the major ity of which were flakes (61% or n = 167). This high flake percent age is Subsurface contexts ac com pa nied by a rel a tively high to tal num ber of A total of 15 units (compris ing eight tren- flake cores (n = 34). Blades and blade cores are ches) was exca vated in Local ity 20, most of which present, though rare in overall assem blage propor - were exca vated to bedrock. On the basis of their tion. Several cores show radial flaking patterns on strati graphic dif fer ences and vari able land scape dorsal and ventral surfaces and some show bidi- contexts, the exca va tion units and their subsurface rectional flak ing. Only one lami nar flake core was arti facts were subdi vided into three main groups: recov ered along Transect 1, the piece showing a northern core, a southern core and a foothills par allel arises along its cen tral ridge. The most zone (Fig. 23). Lithostratigraphic dia grams for the 146 M. Petraglia et al.

northern and southern cores are provided in Fig- ures 24 and 25. Two OSL deter mina tions have been obtained from sedi ments in the northern core, dating to a mini mum of 21 kyr BP (JLP 20B-215) and to ~34 kyr BP (JLP 20B-60). How- ever, these ages can not yet be de fin itively related to other parts of Local ity 20 and further dates are re quired to clar ify their sig nif i cance. The northern core produced a small sam ple of 50 arti facts, which, on the basis of stratig ra phy, can be divided into a lower (NC lower; n = 40) and upper (NC upper; n = 10) collec tion. In contrast, a to tal of 430 ar tifacts was recov ered from the south ern core. These ar tifacts may be di vided into three subsurface subdi vi sions; a lower assem - blage (SC lower; n = 187) in a matrix of silty clay that overlies weath ered bed rock, a mid dle as semblage (SC middle; n = 192) that is sepa rated from the former by a gravel lens, and an upper assem - blage (SC upper; n = 51) consist ing of a rela tively poor con cen tra tion of ar ti facts. The foot hills zone (SF) com prises two rel a tively shal low ex ca va - tions where bedrock was reached af ter ~23 cm in unit 4 and ~90 cm in unit 3. A total of 95 arti facts was retrieved from these foothill exca va tions, Fig. 23. Plan of Jwalapuram Lo cal ity 20 ex ca va tions, how ever, dif fer ences in stra tig ra phy pre vent re li- in di cat ing the lo ca tion of the northern core, southern able in te gra tion of these ar tifacts within the three core and foot hill groups southern core assem blages, though they are com - po nents of the sur face as semblages described flakes, flaked pieces and tab u lar pieces. The foot- above. hills as semblage also has a rel atively large num- The vast major ity of ar tifacts from Local ity ber of cores accom panied by numer ous retouched 20 are made from limestone (97%) and the re- items, in di cat ing the gen eral sim i lar ity of the near main der are made from other raw mate rial types sur face and subsurface as semblages. such as quartz ite, dolerite, quartz crys tal, chal ce- The major ity of the cores ex cavated from Lo- dony and chert. cality 20 are single platform (n = 29 from a total of The NC lower and SC lower as semblages are 66). Multiplatform cores are the next most com - pre dom i nately made up of re touched tab u lar pie- mon (n = 16), fol lowed by bidirectional cores, bi - ces (60% and 49% respectively)4; a few cores, polar cores and a uni direc tional core. Multiplat- flakes and retouched flakes form the remain der of form cores are found in all as semblages ex cept for these col lections. The small NC up per as semblage NC upper, where no cores were found. Bidirec- is com posed entirely of flakes, of which several tional cores were exca vated from SC lower and are re touched, and both unretouched and re- SC middle and bipo lar cores from SC lower and touched flakes numer i cally dom inate the SC mid- the foothills. dle, upper and foot hill as sem blages. Sev eral dif fer ent re touched flake types were The SC has vary ing frequen cies of ar tifact recov ered and are found in differ ent frequen cies types in lower, middle and upper levels, though in all six as semblages. Side re touched flakes are their tech no log i cal char ac ter is tics are gen er ally the most com mon, followed by side and end re- simi lar. Large numbers of cores and as so ciated touched flakes, double side retouched flakes, end flake prod ucts were retrieved, and contain a rel a - retouched flakes and double side and end re- tively high number of retouched items made on touched flakes. One notched Levallois flake was Investigations in the Kurnool Dis trict, South India 147

Fig. 24. Com pos ite strati graphic di a gram of the north ern core group, in di cat ing the lo ca tion of ar ti facts (ar ti fact counts in cir cles) in the NC lower and NC up per groups. The strati graphic po sitions of re cent OSL dates are also pro vided

Fig. 25. Com pos ite strati graphic di a gram of the south ern core group, in di cat ing the lo ca tion of ar ti facts (ar ti fact counts in cir cles) in the SC lower, SC middle and SC up per groups 148 M. Petraglia et al.

re cov ered from SC lower. Blades com prise a very of pre pared cores, a va ri ety of unifacial forms, and small per cent age of the Lo cality 20 ar tifacts, and rare blades and microliths. In spite of the occur - micro liths are simi larly un common, where all are rence of the lat ter amongst the subsurface ar ti- unretouched and found in SC lower, SC middle facts, the Local ity 20 as semblage as a whole can and NC up per (the latter two arti facts are clas si- be de fined as later Mid dle Paleolithic. fied as microblades). Although the northern and southern core as- JURRERU RIVER VAL LEY sem blages do dif fer, there ex ist more sim i lar i ties ROCKSHELTERS be tween these than there are sta tis tically sig nif i - cant dif fer ences. In partic u lar, there are a number The northern slopes of the Jurreru River Val - of simi lar ities be tween the NC lower and SC ley are formed by a quartzitic escarp ment overly - lower groups, and to a slightly lesser extent being limestone beds. During slope retreat, large tween NC lower and the foothills arti facts. In con - quartzitic boulders detached from the quartzite trast, there are sev eral note worthy dif fer ences that units, com ing to rest on the slopes to form natu ral ex ist within the southern core assem blages. Foot - shel ters towards the base of the hill (Fig. 26). Five hill flakes are sig nif icantly heavier than those in rockshelters with ev i dence of hu man oc cu pa tion all three southern groups, whereas there are no were identi fied on the slopes, each with rockart differ ences in weight between the southern and surface scatters of microliths, groundstone groups. SC upper flakes are signif i cantly flatter tools, and pottery. Test exca va tions were con- that those from SC mid dle. SC mid dle flakes pre - ducted to further discern the nature of the stra tig - serve both more cortex and a lower dorsal scar raphy and occu pa tions of these sites. count when compared to all other assem blages (includ ing the northern group). This suggests that, Jwalapuram Lo cal ity 11 as a whole, those flakes in SC middle rep re sent a The Jwalapuram Local ity 11 rockshelter is rel atively earlier stage in the core re duc tion pro - formed by one of the largest boulders laying on cess when compared to the other as semblages. the northern slope. Unfor tu nately, the rockshelter Uni di rec tional flak ing is more com mon than depos its on the eastern side of the boulder were bidirectional or radial flaking when consid er ing exten sively disturbed as a result of modern soil all Lo cality 20 ar tifacts. Larger flakes were se - quar ry ing op er a tions. Al though the de pos its were lected for retouch in SC lower, SC middle and largely removed, remnant stratigraphic profiles foot hills as sem blages, yet par tic u lar flake sizes were pres ent and ex am i na tion of crev ices be neath were not favored in any other groups. The geo- the boul der faces indi cated the possi bil ity for ar- metric index of unifacial reduc tion (Kuhn, 1990) chae o log i cal de pos its. is signif i cantly higher for SC middle and foothills In one crev ice along the south eastern face of flakes when com pared to SC lower and SC upper, the boul der ar chae o log i cal ma te rial was erod ing although there is no differ ence between these from a shallow de posit. The top layer (Level 1) groups in the invasiveness of retouch when apply - was subdi vided into three lateral units. In the ing the index of invasiveness (Clarkson, 2002). western unit, disarticulated and highly fragmented As borne out by these detailed arti fact studies, human bones, proba bly from an adult, were re- Lo cality 20 was a lo cus of raw mate rial pro cure - covered together with pottery sherds, lithic arti - ment and reduc tion. The weathered limestone facts, and small pieces of ochre. Skel etal disarti- clasts were clearly tar geted for core-flake pro duc - culation and high bone fragmen ta tion likely indi - tion. The high inci dence of cortex on cores and cate second ary burial processes prior to the inter - debitage indi cate that pri mary flaking occurred on ment. The central unit of the top layer yielded a site. However, there is also a high frequency of re- dis turbed ju ve nile skel e ton and pot tery. The touch in all Lo cal ity 20 as sem blages, par tic u larly bones appear to be much less fragmen tary than when com pared to other Middle Paleolithic sites those of the adult, and these may have repre sented in the val ley (e.g., Local ities 3 and 21). This is a nearly com plete skel e ton. The un der ly ing unit per haps in dic a tive of rel a tively high oc cu pa tion consisted of a two-meter deep deposit with shal - in ten sity at Lo cal ity 20. The as sem blage con sists low, dark brown sandy sedi ment and numer ous Investigations in the Kurnool Dis trict, South India 149

Fig. 26. The quartz ite boul der forming the rockshelter at Jwalapuram Lo cal ity 9. Note the large size of the boul der based on the in di vid u als stand ing here. This rockshelter has rock art and strat i fied de posits, with an abun dant microlithic assem blage, symbolic pieces, and fauna quartzite rocks. The de posit contained potsherds, Jwalapuram 9 (see below), with a predom inance cattle bone and tooth fragments, and one large, of black-slipped and red-slipped burnished sur- pres sure-flaked blade core of black chert. The faces as well as grey-slipped and burnished sherds vertebrate assem blage (n = 396) con tains many that sug gest conti nu ity from the ear lier thick- reptile bones (perhaps from a single snake), and walled Neolithic pots of this region. The burials in fre quent re mains of do mes tic (two cat tle teeth) are prob a bly as so ci ated with the early/proto- and wild (10 hare remains and three langur re- Megalithic phase, per haps from ca. 1200 BC and mains) ani mals. A fair number of landsnails were their presence suggests that other intact burials also pres ent (n = 95) as well as a sin gle, fresh wa - may be located in undis turbed crevices along the ter bivalve. This deposit rested on top of quartzite per im e ter of the boul der. Dis turbed lime stone spalls and a compact cal crete. A to tal of 96 slabs along the boul der surfaces also indi cate that ceramic sherds from Level 1 was studied, with other Megalithic buri als may be pres ent5. more than a quarter of these badly abraded. Amongst the rest are a great many thin-walled, mainly hand-made, sherds that suggest paddle- Jwalapuram Lo cal ity 12 and-anvil construc tion. Wheelmade or wheel-fin- Local ity 12 is a large boul der provid ing shel- ished sherds account for only about 2% of the as- ter along its west ern side, where a rel atively wide, semblage, which is oth er wise simi lar to that from flat ter race is present. A large crev ice formed 150 M. Petraglia et al.

within the west face of the boulder, creat ing a Jwalapuram Lo cal ity 9 cave-like structure approx i mately 1 m in height Jwalapuram Local ity 9 is a large rockshelter and 1 m in width and extend ing into the boulder in a central and low-ly ing posi tion along the for at least 10 m. The inte rior walls of the crev ice hillslopes. A dense concen tra tion of surface arti - were blackened by soot from fires. Appar ent on facts was present on the southern side of the boul- the surface of the ter race and the hillslope was a der, extend ing more than 25 m from the shelter mod er ate-den sity ar ti fact scat ter, con sist ing pri - wall (Fig. 26). A block of 12 units was placed at marily of microliths with oc ca sional finds of the base of the rockshelter. Two of the units ground and bat tered stone. Several limestone reached a depth of 3.15 m (Fig. 27), and seven dis - slabs were in disturbed posi tions on the terrace tinct strati graphic lay ers were iden tified. Stra tum sur face, per haps orig i nally as so ci ated with Mega - B contained two pit fea tures with ce ramics, lithic burials. whereas Strata C–F are pre-ce ramic lev els with A sin gle test unit was placed on the ter race in abun dant stone ar tifacts. A se ries of cal ibrated ra - front of the crevice entrance during the 2003 sea- diocar bon sam ples date the site from 34 kyr BP son. Stratum A contained pottery, bladelets, towards its base to 12 kyr BP in layers below ce- flakes, as well as char coal and shell frag ments. ramic-yield ing ho ri zons. Stra tum B consisted of a large concen tra tion of Fea ture B1 is a pit con tain ing lithic and ce - quartzite clasts whereas the lower half of the stra - ramic ar tifacts, and was filled with quartz ite spalls tum contained larger boulders. Stratum B, which and limestone slabs. Fea ture B2 was at the rear of was devoid of ceram ics, contained the major ity of the shel ter and con sisted of a pit that was filled the stone tools, in cluding micro liths. Stra tum C with compact sandy-silt sed iment. Fea ture B1 con sisted of a cal cium carbon ate rich de posit, contained a moder ate num ber of arti facts and the which also contained microliths, although the fre- re mains of what ap pears to be partly collapsed quency of the arti facts was lower in com pari son to stone cairn struc tures, per haps in as so ci a tion with the up per strata. Ex cava tion was ter minated at a human burials. Large pieces of burnt or miner al - depth of 1.1 m, though it is proba ble that the ar- ized human bone and ceramic sherds were found chae o log i cal de pos its ex tend deeper, as in Jwala- in close ver ti cal and hor i zon tal prox im ity to these puram Local ity 9. Human bone, consist ing of stone struc tures. Feature B1 overlies B2 and the three hand pha langes, and three possi ble arm bone fill ap pears to be at least partly comprised of new frag ments, were also re cov ered from Stratum B. and old arti facts and rubble that was perhaps ob- Only a very mod est fau nal as semblage (43 tained from the slope in front of the shelter. ver tebrate re mains and 85 mollusc shells) was re - More than 15,000 stone arti facts were recov - covered. In contrast to Jwalapuram 11, no remains ered from the three deepest units. A total of 23 of microvertebrates were found. Only 8 teeth cat e go ries of ar ti facts are rep re sented, in clud ing could be more spe cif ically iden tified; they come forms that re sult from distinc tive kinds of flaking, from small-sized ungulates (includ ing sheep- breakage and heat damage. Ar tifact dis card rates goat-ga zelle), me dium-sized ungulates (in clud ing vary signif i cantly over the depth of the site. a deer), and porcu pine. There is lit tle basis for in- A pronounced peak in stone arti fact discard cen- ter pret ing the taphomomic history of the assem - tered on around 150 cm depth in Stra tum C, with blage. Some remains, in par ticu lar the land- some smaller subsid iary peaks on ei ther side of snails, may have been intro duced as burrow the cen tral one. The bottom most ar tifacts are deaths. The fresh wa ter molluscs and macroverte- found at a depth of 280 cm, around 35 cm above brates were cer tainly intro duced by other agents, the base of the ex cava tion. A smaller dis card peak of which humans seem most likely even though occurs beneath the main peak at around 210 cm we have lit tle di rect ev idence of their in ter ven tion. depth in Stra tum D, and another at around 50 cm The taphonomic history of the human remains is depth in the rubble filled pit feature of Stra tum B1 equally unclear. Jwalapuram 12 contrib utes to the which appears to contain many old arti facts gath - emerg ing pat tern of an as so ci a tion be tween disar- ered up from the slope in front of the shelter. Arti - ticulated human remains and rockshelters with fact numbers in Feature B2 closely mirror those at rock art. equiva lent depth in Stra tum C. Investigations in the Kurnool Dis trict, South India 151

Fig. 27. Stratigrapy of Jwalapuram Local ity 9. Note the burial pits at the top of the section. A stratigraphic hia tus occurs be tween Strata C and D

A se ries of broad tech no log i cal/ty po logi cal cm and decline in abundance fairly rapidly, changes are noted to take place over the span of though they continue to make up a small propor - human occu pa tion. There appear to be a series of tion of the assem blage throughout the rest of the shifts in em pha sis on var i ous tech no log i cal pro ce- de posit. Back ing of small flakes to create sym - dures. Retouch ing is most preva lent in the earli est metric and asym metric forms makes its first ap - pe riod of oc cu pa tion, rep re sented in sed i ments pearance at a depth of around 240 cm, just above below 250 cm in depth, followed by a rapid de - the decline in retouch ing and the decline in the cline with de creas ing depth. Blades greater than 2 fre quency of larger blades. Back ing re mains at cm in length (as opposed to microblades defined high fre quency until just before the peak in ar ti - as be ing shorter than 2 cm) also peak below 250 fact dis card takes place be tween 130–180 cm as 152 M. Petraglia et al.

Fig. 28. Lithic tech no log i cal changes, Jwalapuram Lo cal ity 9

shown in Figure 28. Microblades also make their tion to the Megalithic period (Allchin, 1960; first ap pear ance at this time, and simi larly per sist Boivin et al., 2005; for dating, see Fuller et al., at high frequen cies until around the time total arti - 2007). There is a com plete ab sence of Patpad fact dis card rates peak. Burins make their first ap - Ware Neo lithic types. This suggests a gen er ally pear ance alongside microblades, but peak later post-Neolithic date but perhaps earlier than the during the period of greatest arti fact discard. This classic mega lithic when wheel fin ished types may sim ply be a by prod uct of sam ple size ef fects, would be expected to domi nate. The profile of however, as com monly occurs when rare items wares (Fig. 29) when compared to the Neolithic are only found in larger as sem blages. Cores of cave de pos its in di cates a dif fer ence in em pha sis vari ous kinds then rise to promi nence once burins on burnished black-slipped types, which is more begin to decline. Finally, the use of high quality in keeping with an Iron Age date. A small num ber chalce dony rises to promi nence in the last 150 cm, of red-slipped polished sherds are indic a tive of a peaking in the last spit. Arti facts of worked bone, Mega lithic time ho ri zon. A few rec og niz able shell, and other mate ri als were recov ered through- forms are also of Iron Age types, includ ing an out the Local ity 9 deposit, includ ing worked iron open bowl form with inturn rim (Allchin, 1960, in the upper levels and a num ber of bone and shell Form Class B2), and a narrow-neck evert rim jar beads as so ci ated with the mi crolithic ar ti facts. (Allchin, 1960, Form Class B25). Both of these Ceram ics came only from the up per feature forms can be regarded as evolving from Neolithic B1, which appears to have been a burial pit. A to - an teced ents (Form Classes A7 and A27, re spec - tal of 983 sherds were recov ered and exam ined. tively) and to occur rather ear lier than more elab o - Most sherds are thin walled, and might have been rately rimmed jars and carinated bowls which are made by paddle-and-an vil technique rather then more typ ical of the Clas sic Megalthic pe riod. the coil methods consid ered more typi cal of Neo- Taken as a whole, and in contrast to open air clas- lithic ceram ics. Among these a small per cent age sic mega lithic sites, such as Mogasarayanigondi I (6.5%) show evi dence of rilling suggest ing that (see below), this site can be suggested to date to they were made, or at least finished, on a wheel or the early part of the Iron Age, which in the turn table, a tech nique that came into use at the Kurnool District can be suggested to start from ca. very end of the Southern Neolithic and the transi - 1200 cal. BC (Fuller et al., 2007: Fig. 6), contin u - Investigations in the Kurnool Dis trict, South India 153

Fig. 29. Jwalapuram Local ity 9 rep resen ta tion of ceramic wares based on to tal sherd counts. Smaller sherd frag - ments may over es ti mate the sig nif i cance of a given ware. Sherds are domi nated by slipped bur nished types, mostly hand made but thin-walled. This cou pled with the pres ence of small pro por tions of red-pol ished ware sug gests a date towards the early 1st millen nium BC and the tran sition from the Late Neo lithic to the Iron Age ing to per haps the mid-first millen nium BC. As of several gen era (e.g. Gazella/Tetracerus). Con- such we might regard burials within these rock firmed re mains of do mes tic an i mals (horse/ass, shel ters as early “megaliths” that took ad van tage buffalo, buf falo/cat tle, goat) were ex tremely rare of nat urally prom inent stones as markers. That and restricted to phases B1 and B2 (megalithic this site might have been visited over some period age). Re mains of wild an imals were pres ent of time during the Megalithic period is suggested throughout the se quence. The small microverte- by an appar ent trend in which wheelmade sherds brate as semblage (NISP = 63) in cludes a diverse are somewhat more fre quent to wards the sur face. range of taxa includ ing rodents, birds, reptiles, Jwalapuram Local ity 9 pro vides a rela tively amphib ians, and fish. Microvertebrates are rare large faunal assem blage of 2732 non-human ver- throughout the se quence, never account ing for tebrate remains and 1644 mollusc shells. To these more than 3% of the as sem blage. totals we can add a further 24 fragmen tary human The rel atively large mollusc as semblage (n = re mains. Of the 2669 macrovertebrate frag ments 1306 shells) is in some ways more inter est ing than pres ent, only 8.6% were iden tifi able. At least 19 the ver te brates. First, they show that the col lect ing taxa are present, includ ing the fol lowing: horse/ of molluscs, par tic u larly fresh wa ter bi valves (n = ass (Equus sp.), buffalo (Bubalus bubalus), goat 381), was not a minor or inci den tal practice, but a (Capra hircus), pig/boar (Sus sp.), nilgai (Bosel- sys tem at i cally used strat egy dur ing all phases of aphus tragocamelus), blackbuck (Antilope cervi- site use. A sig nif icant de crease in bivalve size capra), ga zelle (Gazella gazella), four-horned an- over time (between phases C and B2) may have te lope (Tetracerus quadricornis), sambar (Cervus been caused by sustained preda tion pressure (Fig. unicolor), chital (Axis axis), muntjac (Muntiacus 30). Large landsnails (Cryptozona sp.) appear to muntjak), pangolin (Manis gigantea), jungle cat have been collected and consumed during the ear- (Felis chaus), a dog/wolf/jackal (Canis sp.), mon - lier phases (D, C, and per haps B2); they dropped goose (Herpestes edwardsi), black-naped hare out of the diet during the later phases of occu pa - (Lepus nigricollis), hedge hog (Erinaceus sp.), tion. The smaller landsnails and the freshwa ter langur (Presbytis entellus), and porcu pine (Hys- snails proba bly accu mu lated at the site during pe- trix in dica). Many “iden ti fi able” re mains were ri ods of less in ten sive hu man oc cu pa tion – their only identi fi able to body-size cate go ries or to one frequency varies inversely with the frequency of 154 M. Petraglia et al.

Fig. 30. Mean shell size in landsnails (Cryptozona sp., N = 30) and freshwa ter bi valves (Parreysia sp., N = 79, and Cyclotopois sp., N = 18) at Jwalapuram Local ity 9

macrovertebrate remains; the in ten sity of human Jurreru River Val ley. Sur face sur vey re sulted in use of Jwalapuram 9 de creased between phase C the iden tifi ca tion of four large sites with pot tery and B2. Tempo ral shifts in the size of the land- on surface (i.e., Pathapadu I and II, Mogasara- snails (Cryptozona and Cyclotopois) are more dif- yanigondi II, Deyyapu Gundu Habi ta tion). The fi cult to in ter pret, but prob a bly re flect as yet un - Pathapadu I lo cality was lo cated near the classic known en vi ron men tal factors. Neolithic site (e.g., Allchin, 1962; Sarma, 1968). Frag men tary hu man re mains are pres ent Pathapadu II oc curred at the base of the southern throughout the sequence. The few burned cranial hillsopes and was nota ble for its scatter of pottery fragments and isolated teeth from phase C may and chipped stone arti facts. The site occurs below have been scattered from a dis turbed burial, or a distinc tive limestone outcrop, and it appears that they may rep re sent de lib er ate de po si tions/in clu- some of the chipped stone arti facts were produced sions by peo ple vis iting the site. In the lat ter from the same mate rial 6. The sur face col lections event, they might provide clues of ways in which from Mogasarayanigondi II were lim ited and de - peo ple main tained/cre ated links among dif fer ent rived from a loca tion where recently upturned places, peoples, and events. Human remains from soils produced a small number of potsherds, bone the megalithic use of the site are more complete fragments, and chipped stone arti facts. The than those from the earlier occu pa tions and now Deyyapu Gundu habi ta tion site appeared to be re- include only postcranial remains and a few iso- cently plowed, and consisted of a dense concen - lated teeth. Par tic u larly sig nif i cant is a con cen tra - tra tion of ar ti facts, in clud ing large pot sherds and tion of fragmen tary long bones from a single con - chipped and ground stone arti facts. Several mega- text in phase A; these re mains may repre sent a lithic struc tures were found near the Deyyapu sec ond ary inhumation – “nat u ral” mega lithic Gundu habi ta tion site, consis tent with the Iron burial. Age sherds identi fied on-site. Later occu pa tion in the Early His toric pe riod is in di cated by ce ramics and a diag nos tic coin. With the excep tion of the Deyyapu Gundu site, each of the sites was tested Neo lithic to His toric Era Lo cal i ties for subsurface re mains. Two explor atory units Sub stan tial open-air hab i ta tions, burial fea- were placed in each of the three sites, but none of tures and historic structures are present in the the ex plo ra tions pro duced in tact de pos its. Strati - Investigations in the Kurnool Dis trict, South India 155

Fig. 31. Mogasarayanigondi I dol men on the hillcrest. This dol men is ac compa nied by many oth ers on this hillcrest and along the north ern slopes of the valley. The large up right lime stone slabs were ob tained from the sur - round ing hillslopes. These dol mens and other mega lithic structures are be ing ac tively looted. Skele tal re mains, pot - tery and other ar ti facts are commonly found in spoil piles graphic pro files in di cated that the ar chae o log i cal differ ent sizes, though many are rectan gu lar in de pos its were pri mar ily con fined to the up per most shape and contain slabs in an upright posi tion. black clays, which have been subse quently tilled. The limestone that outcrops on-site appears to be Mod ern ex pan sion of ag ri cul tural fields and con- the mate rial used for dolmen construc tion. The struc tion of irri ga tion canals at Pathapadu I appar - sur round ing ag ri cul tural fields con tain a few ex- ently destroyed the habi ta tion mound that was re - amples of the more substan tial cairns and cists, ported by ear lier in ves ti ga tors. but the great major ity of these structures appear to The Jurreru River Valley and its envi rons have been dis assem bled by farm ers. Prior looting contain a large number of megalithic burial struc - of the dol mens has revealed frag men tary skel etal tures (e.g., Mogasarayanigondi I, Deyyapu Gundu el e ments, cre mated hu man re mains in terred in ce - hab i ta tion, Jwalapuram Lo cal ity 14). Al though ramic wares, and burial goods such as Iron Age par tic u lar lo ca tions have been rec og nized, the ceramic pots, jars, and bowls and copper rings. mega lithic struc tures form an ae ri ally ex ten sive Cleaning of the spoil piles of one of the looted landscape and occur along many of the hillslopes dol mens re vealed the pres ence of mul tiple ce - on the north side of the valley. Mogasarayani- ramic wares in undis turbed depos its. These con - gondi I is a concen tra tion of at least 25 dol mens sisted of slipped black and red finished types, ei- on a small hillcrest (Fig. 31). The dolmens are of ther burnished or polished to high shine, inclu- 156 M. Petraglia et al.

Fig. 32. Construc tion of modern Hindu shrine at the rockshelter of Katavani Kunta 7. This shrine is an exam ple of the modern use of rockshelters by dev otees

ding many of the black and red ware bowls with The rubble form ing the cairns are made from the carinated pro files and elabo rately rimmed jar local quartzite cobbles and boulders whereas the sherds that are typi cally as so ci ated with the South upright burial cham bers are made from local lime- India megalithic (cf. Wheeler, 1948). The pre- stone slabs. domi nance of these clas sic mega lithic types asso - The Jurreru River Valley and surround ing ar - ci ated with typi cal cist burials sug gests that eas con tain ev idence for Me di eval and Historic Mogasarayanigondi I is a Clas sic Mega lithic cem - pe riod struc tures, in clud ing nearby 12th cen tury e tery from the latter half of the first millen nium tem ple struc tures and the 17th century fort at BC and presum ably later than the burials in the Banaganapalle. The temple at Yaganti is a Saivite rockshelters (Jwalapuram Local ities 9 and 11). tem ple known as the Sri Yaganteshwara Swamy A large num ber of intact cairns and cists oc - Temple or Temple of Uma Maheshwar. It was cur below the rockshelters (i.e., be tween Jwala- built in the 15th cen tury by the first Vijayanagara puram Local i ties 11 and 16). The megalithic kings, and is asso ci ated with a natu ral spring and structures are hid den from view as they are in aca - number of nota ble natu ral limestone caves, all of cia-domi nated landscapes and on hillslopes that which have been incor po rated into the tem ple are cur rently un pro duc tive for farm ing. Jwala- com plex. The presence of the spring and caves puram Local ity 25 is one loca tion where modern sug gests that the lo cal ity likely had a sa cred sig - soil quarry ing occurred, thereby destroy ing a cou - nifi cance prior to the building of the current tem - ple of structures that contained Iron Age ceram ics. ple. The temple at Yaganti is a place of pil grim- Investigations in the Kurnool Dis trict, South India 157 age, with pilgrims pass ing through the Katavani ment in addi tion to the red and white. Three shel- Kunta Valley (tra dition ally on foot) when travel - ters have red or yellow ochre depic tions. The ling from the Sri Maddileti Narasimha Swamy other 13 sites can be dif fer en tiated as rel atively re- Tem ple in Rangapuram vil lage, south of Betha- cent and/or contem porary Hindu sites in the mcherla. The latter is ap par ently a Vaishnavite Katavani Kunta Val ley, many ap par ently re lated tem ple, and abundant Vaishnavite namalu sym - to the ac tiv i ties of pil grims trav el ing a pil grim age bols mark several large boulders in the Katavani route between two tem ples. The 13 Hindu sites Kunta Val ley, above the Yaganti temple complex. pri mar ily in clude de pic tions in white of re li gious Both Vaishnavite and Saivite imag ery is found sym bols, and these show a signif i cant degree of within the Yaganti temple complex, re lat ing to at - super-po si tion ing, with pil grims con tin u ing to tempts to rec on cile the ri val sects dur ing the Me - paint symbols on the rock surfaces up to the pres - dieval period. The hillslopes on the eastern side of ent day. the Yaganti tem ple contain many rockshelters The 27 rockshelters bearing red images con- formed by large quartzite boulders. Survey of tain in total up to 400 identi fiable paint ings on the some of the boulders indi cated that the shelters shelter walls. The number of paintings ranges have been used for reli gious purposes, as indi - from only one identi fi able painting in several cated by paint ing of Hindu symbols on shelter rockshelters to a maxi mum of 79 paintings in a walls and by construc tion of small shrines (Yaga- single shelter. A total of 10 of the 27 shel ters have nti 4, Katavani Kunta 7) (Fig. 32). Archae olog i cal super-posi tion ing, but usually only one or two fig - depos its older than the tem ple complex are pres - ures have been added over previ ously painted im - ent within the shel ters, as shown by microlithic in - ages. Many of the paintings are isolated and the dustries in rockshelters (e.g., Katavani Kunta 1) lack of com pos ite im ag ery seems de lib er ate. Of and by megalithic structures in other sites (e.g., the 27 sites with red im ages, 13 contain archae o - Yaganti Lo cal ity 3). log i cal re mains, in clud ing four shel ters with microlithic ar tifacts, two shel ters with pot sherds, five shelters with both microlithic ar tifacts and ROCK ART AND ROCKSHELTER potsherds, and one with a megalithic cist. LOCALES AND LAND SCAPES The 400 paintings, most of which are solid or A total of 50 rock art sites were located dur- out lined red, in clude an thro po mor phic fig ures ing two sea sons of survey. The rock art sites form (often shown in profile with a bent leg and raised a sig nif i cant fea ture of the ar chae o log i cal land - arm) (Fig. 34), lizard-like therianthropes (surreal scape of the Kurnool District, but have not been crea tures com pris ing both an i mal and hu man fea- tra di tion ally re ported by ar chae ol o gists work ing tures), ani mals (e.g., deer, goats, boars, moni tor in the region. Based on the presence of thousands lizards, car ni vores, an el ephant, a cow, a snake, of unsurveyed boulders in the region, it is proba - fish), geomet ric figures (e.g., V-shapes, sets of ble that there are many hundreds of rockshelters con cen tric cir cles, zig zags, sets of dots, Vishnu with rock art still to be found. With the excep tion sym bols), hands, handprints, pairs of feet, and of a sin gle limestone rockshelter (MCG IV), all con tain ers. Al though scenes are gen er ally rare, rock art is only known from the walls of quartzite one spectac u lar panel occurs at Katavani Kunta 3, boulders. Forty of the fifty rock art sites were draw ing to gether a va ri ety of an thro po mor phic closely in spected and re corded at Jwalapuram, figures, an imals (fish, snake), liz ard-like therian- Yaganti and Katavani Kunta. While most of the thropes, and geometrics. The fish depic tions (to- sites bear soli tary im ages or as semblages of in di - gether with fresh wa ter shell fish in ar chae o log i cal vidual im ages, one of the rock art sites is com - depos its) occur in an envi ron ment that is pres ently posed of a spec tac ular panel (Fig. 33). dry, per haps in di cat ing a change in en vi ron ment. A total of 27 of the 40 rock art sites have im - Sev eral pot-like con tain ers are drawn at Katavani ages in red ochre (ferric oxide) pigment. Of these Kunta 4 – if these are indeed pots, they repre sent 27 sites, nine also have depic tions in white pig - ar tifacts that are rarely de picted in global rock art. ment (kaolin and/or lime), and of these nine bi- Though the rock art is diffi cult to date, it ap- chrome sites, two also have black (char coal) pig - pears likely that the im ages have been produced 158 M. Petraglia et al.

Fig. 33. Rock art panel at Katavani Kunta Lo cal ity 3. The rock art is made with red ochre pig ment. This panel con - sists of an thro po mor phic fig ures, therianthropes, and var i ous an i mals

over a substan tial pe riod of time. Test exca va tions It is clear that the Hindu re ligious symbols are under neath the painted shelter at Jwalapuram Lo- less than ca. 2,000 years old (the begin ning of the cality 9 re vealed more than three metres of de pos - ‘modern’ Hindu reli gion), and many are al most its, the overwhelm ing major ity of which were certainly less than 500 years old (based on our un- aceramic and domi nated by microlithic as sem- derstand ing of the age of the Yaganti tem ple). blages with cal i brated ra dio car bon dates from Such rock art is gener ally distin guish able not only 12–34 kyr BP. Other rockshelters with rock art by its content but also the pigments used to create fea ture sig nif i cant microlithic scat ters, sug gest ing it, which are gener ally thick, white and well-pre - that testing of these other shel ters might yield sim - served, or arti fi cial. White de pic tions have been ilar find ings. It is likely that the shel ters where reported inside dolmens, hence some may be as rock art is found were in use for a long period of old as the Iron Age (K. Rajan, personal comm., time, and it is possi ble that some portion of the 2005). Ethnographic enqui ries also indi cate that rock art is pre-ceramic. It is, of course, impor tant the white depic tions are still being produced to - to bear in mind that ‘tribal’ groups living a forag - day. The vast major ity of this later and contem po - ing lifestyle have long inhab ited the more remote rary rock art re lates to Hindu re ligious prac tices, areas of Andhra Pradesh, and that some of the and includes marks made during puja (wor ship) microlithic lev els might well be his tori cal in date and, in the Katavani Kunta Valley, marks made by (or indeed Neolithic or Iron Age). pass ing pil grims. Investigations in the Kurnool Dis trict, South India 159

Fig. 34. Ex ample of rock art image, anthropomorph over deer at Katavani Kunta Lo cal ity 1

DIS CUS SION Pappu and Deo, 1994; Petraglia, 1998, 2006; Korisettar, 2007). The se lec tion and use of quartz- The Kurnool District contains a range of ar- ite and limestone for stone tool manu fac ture com - chae o log i cal sites that pro vide an op por tu nity to ple ments findings in nearby ar eas. These raw ma- obtain new infor ma tion about the history of hu- te ri als were pro cured in the site vi cin ity, from man activ i ties in southern India. The two seasons readily avail able clasts, such as grav els or weath- of ex plor atory re search rep re sent a rel a tively ering out crops. The tools are of vari ous types, small ef fort com pared to the large ar chae o log i cal sizes, and shapes. The LCTs of Jwalapuram Tank po ten tial of the re gion. Nev er the less, our re cent have not been transported from their origi nal con - work provides some insights into regional prehis - texts and occur along paleofans at the base of tory that will form the basis of more de tailed study hillslopes and in palludal contexts. Their shallow in the near future. The fol lowing discus sion pres- burial and lo cal ized re work ing, how ever, pre- ents some working hypoth e ses based on current clude be hav ioral in ter pre ta tion from a spa tial per- in for ma tion to date. spective. Based on the presence of long sedimen- The most an cient de pos its with ar chae o log i - tary pro files in the Jurreru River Val ley (e.g., at cal asso cia tions are seen at the Jwalapuram Tank. the Jwalapuram Dry Well), it is proba ble that The Large Cutting Tools found at this local ity are Acheulean as semblages are also deeply bur ied un - typi cal com ponents of the Acheulean technocom- der neath al lu vium. plex, which is found in other basins across south- The iden ti fi ca tion of well-pre served tephra ern India (e.g., Paddayya, 1982; Raju, 1988; depos its in the Kurnool District is of major signif - 160 M. Petraglia et al.

i cance to our po ten tial un der stand ing of re gional ity 21, which has been dated to ca. 38 kyr BP, con - envi ron men tal changes and its im pact on hominin sists of 176 ar tifacts, with the major ity (n = 82) pop u la tions. Al though tephra de pos its have pre vi - concen trated in the upper 20 cm. Exca va tions at ously been found in the subcon ti nent (e.g., Acha- Local ity 17, the most northern exca vated ash site, ryya and Basu, 1993; Williams and Clarke, 1995), pro duced a small assem blage of 37 ar tifacts. A to - their as so ci a tion with ar chae o log i cal ma te ri als has tal of 575 ar tifacts were recov ered from sev eral re mained con tro ver sial (e.g., Mishra and ex ca va tions at Lo cal ity 20 and the ma jor ity of Rajaguru, 1994, 1996; Jones, 2007). The deposi- these are from the southern core area (n = 430) tional posi tion of the tephra in the Jurreru River with rela tively few from the northern core (n = 50) Valley, together with new chronometric and geo- and southern foot hills (n = 95). chemi cal ev idence, in di cates that the ash is a pro- Con sid er ing all cores re cov ered from ex ca va- duct of the 74,000 year old YTT event (Petraglia tions at local ities 3, 17, 21 and 20 (exclud ing the et al., 2007). The recov ery of Middle Paleolithic foot hills as semblage), multiplatform cores are the assem blages in dated de posits below and above most common (n = 22), followed by single plat- the ash provides the first firm archae olog i cal asso - form cores (n = 18), bidirectional cores (n = 4), bi- cia tion, helping to resolve some of the uncer tainty polar cores (n = 2), unidi rec tional cores (n = 1) about ar chae o log i cal as so ci a tions in In dia. The re- and microblade cores (n = 1). Levallois and blade covery of Middle Paleolithic indus tries without cores are ab sent. Plat form prep ara tion is pres ent ma jor tech no log i cal changes af ter the ashfall, in - in the form of overhang removal and facet ing, di cate that pop u la tions sur vived the super-erup- where the former is consid er ably more com mon. tion. We have suggested, based on a va riety of The cores show 32% overhang removal, 6% fac - datasets, that modern humans were present in the eting, 13% a com bina tion of overhang removal In dian sub con ti nent be fore the super-erup tion, and facet ing and 49% of cores show no prepa ra - and that these popu la tions survived this event tion. Flakes pre serve ev i dence of prep a ra tion; (Petraglia et al., 2007). This is not to suggest that 33% exhibit over hang removal, 5% are faceted, there were not ef fects on envi ron ments and popu - 1% show both and 61% show none. Blades are lations as a result of the Toba super-eruption, present but rare, compris ing 2.1% of all arti facts how ever; in fact, field ob ser va tions in di cate that compared to 7.2% at Lo cal ity 9. Micro liths are the tephra marks some changes in the imme di ate even rarer, mak ing up 0.9% of all ar tifacts, com- site envi ron ment. We have argued that the ponded pared to 5% at Local ity 9 (n = 64, of which 44 are envi ron ments that existed prior to the ashfall were backed). No retouched microliths are encoun tered no longer present after the event. Indeed, the de- in the Mid dle Paleolithic as semblages. Burins are posi tion of thick ash units in the central val ley also rare, mak ing up 0.3% of all ar tifacts, com- would have clogged river channels and polluted pared to 1.1% at Lo cality 9. However, sev eral dif - fresh wa ter sources. Hence, it is prob able that ferent types of retouched flakes form a signif i cant landscape al ter ations would have presented adap - part of the Mid dle Paleolithic as semblage, clas si - tive dif fi cul ties for hu man pop u la tions, as plant fied accord ing to the loca tion of retouch. Retouch and ani mal com muni ties would have been nega - is char ac ter is ti cally in for mal and un stan dard ized. tively im pacted by geomorphic changes in the val - Of all Mid dle Paleolithic ar tifacts ex ca vated, ley. Alter ations in tool type fre quen cies and 15.4% are retouched flakes (n = 129, of which n = changes in raw mate ri als may be signs of some 27 are notched). Side retouched flakes are the adaptive differ ences in pre- and post-Toba situ a - most common form, making up 4.3% of all arti - tions. facts. These are followed by side and end re - The Middle Paleolithic is well repre sented in touched flakes (2.8%), double side and end re- the Jurreru River Val ley. Lo cal ity 3 has yielded touched flakes (2.5%), end re touched flakes well-dated Mid dle Paleolithic as semblages, be - (2.2%) and double side retouched flakes (1.9%). neath and above the ash, at 77 and 74 kyr BP, re- Other forms of retouched flakes only consti tute a spec tively. The Lo cal ity 3 as sem blage con sists of small pro por tion. 323 arti facts, where 215 were recov ered from be- Two other char ac ter is tics of the Mid dle neath the ash and 108 from above the ash. Local - Paleolithic of the val ley are im por tant to note. A Investigations in the Kurnool Dis trict, South India 161 piece of red-ochre which shows signs of use was cm) at the base of the 3 m section, datable to re cov ered dur ing ex ca va tions be low the ash at Lo- 34,000 cal i brated ra dio car bon years and older. cality 3. Sec ond, re touched tab u lar pieces make While it is proba ble that the caves contain blades up 15.2% of all arti facts and are found al most ex - of Late Pleis to cene age, larger samples sizes and clusively at Local ity 20. They oc cur in the lower, ra dio metric dates are needed to deter mine how middle and upper levels of the exca vated depos its these cave indus tries fit into an overall techno log - and therefore rep re sent a tra di tion that persisted ical sequence. More over, though many of the pre- for some time. The predom i nance of this type at viously identi fied worked bones may be the result Local ity 20, where sources of limestone from of ani mal contri bu tions, the range of fauna in the which they are made are abundant, suggests a caves is im portant for Indian prehis tory, as firm “least-cost” or op por tu nis tic be hav ior: re touch ing ev i dence for Late Pleis to cene fau nal as sem blages small tabu lar pieces for the purpose of obtain ing a has been lacking. suitable cutting or scraping edge is less labor-in - Lo cal i ties with microlithic in dus tries form the tensive than produc ing flakes for retouch via core most abun dant sources of ar chae o log i cal ev i dence prepa ra tion and reduc tion. The fact that these in the region. Current research indi cates that types persist throughout the depth of the Local ity micro liths are cer tainly as so ci ated with ce ramic 20 depos its, and that this behav ior was therefore as sem blages (e.g., Jwalapuram Lo cal ity 9, not tem po rally iso lated, is par tic u larly in ter est ing. Sanyasula Gavi, Caves 1 and 2), though we have The Middle Paleolithic of the Jwalapuram observed that core types in later indus tries can be val ley con sists of core-flake tech nol o gies with char acter ized as fluted cores. Microlithic in dus - atyp ical scrap ers as well as notched flakes form- tries, how ever, pre-date ce ram ics, ex tend ing into ing the major ity of re touched types. Burins are the Early Ho lo cene and Late Pleisto cene. Abun - rare, as are blades and microliths. Bipo lar tech nol - dant microlithic evi dence was de rived from the ogies exist but again, these are uncom mon. In strat i fied de pos its of Jwalapuram Lo cal ity 9, one con trast, the ar chae o log i cal de pos its of Lo cal ity 9 of the best strat ified microlithic sites in In dia. show a no table in crease in microblade cores, AMS dat ing of microlithic lev els in di cates a cal i - blades (and the ap pear ance of blade cores), micro - brated age range from 12 to 34 kyr BP. The age of liths (and microblades) and the first occur rence of Jwalpuram 9 is therefore compa ra ble to the Sri backed vari et ies of these. Conse quently, the fre- Lankan caves where microliths range from the quency of retouched flakes declined (compris ing Late Holo cene to 28.5 kyr BP (Deraniyagala, only 2% of the Local ity 9 as sem blage, in contrast 1992). Microlithic assem blages at Jwalapuram 9 to 15% for the Middle Paleolithic) as backing be- are as so ci ated with dec o ra tive pieces, such as came more com mon place. beads and worked bones, includ ing har poon and Though the Kurnool District is famous for its possi ble awl fragments. Though no dates are cur- Upper Paleolithic assem blages, it re mains diffi - rently avail able for the pro lific rockart at cult, based on our current research, to verify the Jwalapuram 9 and other rockshelters, it is possi ble ex is tence of clas sic tool types nor mally as so ci ated that some of the paintings extend into the Late with such an indus try, such as pris matic blade Pleis to cene. The re cov ery of hu man cra nial re - cores, lam inar blades, blade tools (e.g., burins, mains, as so ci ated with a cal i brated ra dio car bon endscrapers) and worked bones. Our lim ited test date of between 12 and 14 kyr at Jwalapuram 9, is exca va tions in MCG and Chapter House North in trigu ing, sug gest ing the po ten tial for hominin Cave produced low arti fact counts, mostly flakes re mains from rockshelters and caves. The thick - and chips with rare finds of blades, though Murty ness of the cra nial re mains may in di cate the pres - (1974) indi cated that blades were plenti ful in the ence of humans with robust skele tons, compa ra - stone tool assem blages (41%), and Thimma ble to those of the Early Holo cene in the Ganga Reddy (1977), on the other hand, indi cated that Plain (Lukacs, 2007). blades were rare (12%). With re spect to the pres - With respect to diet, a range of ter res trial and ence of blade indus tries, the only obser va tion that aquatic fauna was utilized. The rela tively large we can make is that Jwalapuram Local ity 9 has mol lusc as sem blage, par tic u larly fresh wa ter bi - yielded a higher fre quency of lami nar items (>2 valves, in di cates that col lection of these re sources 162 M. Petraglia et al.

was not a minor prac tice, but a sys tem atic strategy from the Neolithic onwards, but that they were in during all phases of site use. There is a signif i cant exchange with set tled popu la tions for long pe ri- de crease in bivalve size over time, per haps as so ci- ods of time (Murty, 1981, 1985; Morri son, 2007). ated with sustained preda tion pressure. Rock walls at the entrances of caves (e.g., Jinu’s Con tem po rary with open-air set tle ments from Ladder) may indi cate the use of the Kurnool caves the Neo lithic to the Early Historic, the caves by hunter-gather ers, who have been observed to (Chapter House North Cave, Sanyasula Gavi, block off the entrances of caves for porcu pine Caves 1 and 2) and rockshelters (Jwalapuram hunting and trapping (Murty, 1974, 1985). We Localites 9, 12) show evi dence for human activ - suggest that mobile hunter-gather ers, in contact ity. Patpad Neo lithic pot sherds, micro liths and with set tled commu ni ties, used these caves based char coal-bearing lev els were pres ent in Chapter upon the presence of ceramic sherds (repre sent ing House North Cave and the two Sanyasula Caves. only a few ves sels), the absence of other arti facts A quern at Sanyasula Cave 1, an ash bed in Cave typ i cal of ag ri cul tural com mu ni ties, the spo radic 2, and ra dio car bon ages on do mes ti cated mung- nature of site use (includ ing depositional hia tuses bean and wild seeds between of 3,995 to 1487 BP between strata), and the dis tant loca tion of the in di cate plant pro cess ing and burn ing ac tiv i ties in caves from lowland settle ments. Based upon our the Ho lo cene. Of in ter est is the fact that the ear li- re sults, we have mod elled this sit u a tion, in di cat - est dates as so ci ated with Neo lithic de pos its in ing the movement of peoples from lowland to these caves, in clud ing the direct AMS date on highland contexts (Fig. 35). In addi tion to the pos- mungbean from Sanyasula caves (ca. 1900–1800 si bil ity for for ag ers, it is clear that the caves and BC), pre date the es tab lished Neo lithic open air rockshelters have been used by reli gious figures. villages in the Kunderu river valley to the east by Modern habi ta tion is present in the Sanyasula at least one or two cen tu ries (cf. Fuller et al., Caves, where hearths and limestone chipping was 2007). This may suggest that the Neolithic occu - docu mented. Indeed, the marking of rockshelters pations at these caves are im portant for under - with Hindu symbols is a tradi tion that proba bly standing the process of Neolithization in this re - extends back many hundreds of years, espe cially gion. Previ ously, it was proposed that the Neo- in the rockshelters near Yaganti Temple. lithic in Kurnool District might be seen in terms of In closing, it is incum bent upon us to re port a “static frontier” model in which local hunter- that many of the ar chae o log ical sites in the gather ers adopted food produc tion from their es- Kurnool Dis trict are be ing en tirely oblit er ated or tablished Neolithic neighbors to the west (Fuller se ri ously im pacted by mod ern eco nomic de vel op - et al., 2001b; Fuller, 2006: 53). We might now ment. Dam and ir ri ga tion ca nal pro jects, in ten si - consider whether the occu pants of the caves in the fied ag ri cul tural ex pan sion, com mer cial quarry Erramalai Hills were at the fore front of inter ac - oper a tions, and site loot ing are all ac tively con- tions with the Neolithic to the west or could have tribut ing to the alter ation and destruc tion of ar- been “pio neer” farmers who moved into the area chae o log ical sites and the wider land scape. To from the west, with an initial shift to an in creased make matters worse, no her itage manage ment reli ance on gathered and hunted resources. In study has been conducted to cata logue and docu - either case, the occu pa tions at these caves con tin - ment the range of the region’s archae o log i cal ued through the estab lish ment of open air agri cul - sites. In just several years of working in the tural set tle ment fur ther east, char ac ter ized by Kurnool District, we have witnessed the de struc - Patpad Ware from ca. 1700 cal. BC. tion of a num ber of sig nif i cant ar chae o log i cal Much of the Holo cene fauna from the caves sites, and have noted that many more are under also appears to have been hunted, despite the pres- threat. Com mer cial lime stone quar ry ing threat ens ence of some livestock. Taken together, such evi - the Billasurgam caves, and soil mining has al- dence in di cates the use of caves and rockshelters ready de stroyed some rockshelters, such as Jwala- by for ag ers who were con duct ing ac tiv i ties in puram Lo cal ity 11. Ag ri cul tural ac tiv i ties ap pear highlands, some dis tance from set tled vil lages. to have obliter ated the famous Pathapadu Neo - His tor i cal and ethnographic in for ma tion in di cates lithic site and its habi ta tion mound. Construc tion that for ag ers were in creas ingly cir cum scribed of rice paddy fields has completely erased some of Investigations in the Kurnool Dis trict, South India 163

Fig. 35. Hy po thet i cal for ag ing range of hunter-gath er ers in the Kurnool Dis trict. Ar rows de pict for ag ers mov ing among the caves in up lands. Other ar rows repre sent trans fers be tween for ag ers and agropastoralists set tled along the Kunderu river

our newly re corded ar chae o log i cal sites in the Notes Jurreru River Valley. Road construc tion, agri cul - 1. We con ducted limited sur vey in a tun nel-like tural field de velop ment, and loot ing have contrib - pas sage in side Purga tory Cave. The pas sage was nar - uted to the destruc tion of megalithic structures row (1–2 m in width, 1 m in height) though long (ex- and even en tire Mega lithic ceme ter ies. In an at- tend ing for >0.5 km), re quir ing crawl ing in places. A tempt to com bat this adverse sit ua tion, we have few arti facts (i.e., a hammerstone, historic period ce- ini ti ated a pub lic out reach and ed u ca tion pro gram ramics) and fresh bone were collected on the surface of the floor. that aims to raise awareness con cern ing the re - 2. The au thors of the pub li ca tion re ported find ing gion’s ar chae o log i cal re cord. We have held dis- cra nial and den tal frag ments of a “hominid” in the cave cus sions with vil lage head men, vil lag ers, and ex ca va tions. No age was in di cated for these hu man re - school chil dren, highlight ing the region’s rich cul - mains. Our in spec tion of the Mogasarayanigondi cave tural her itage. We have be gun to conduct inter - and nearby ones in di cated that they were un likely liv - views with local and regional newspa pers and we ing ar eas as they were very high up from the val ley have de veloped a brochure (in English and Tel- floor and the caves them selves were small open joints ugu) address ing the archae ology of the local area. in the lime stone. Given that Iron Age buri als are com - In light of the real ity of the rapid pace of eco- mon on the val ley floor, we sus pect that the “hominid” bones are not an cient, and prob a bly date to the Ho - nomic devel op ment, and the lack of im plemen ta - lo cene. tion and enforce ment of cultural resource regu la - 3. Sur vey of the area did not iden tify the Patha- tions, it is appar ent that public outreach and padu hab i ta tion or any as so ci ated fea tures. Neo lithic edu ca tion is a way to help protect the region’s rich arti facts were spread across agri cul tural fields in the vi- cultural heritage. cinity of the site, but no intact depos its were identi fied 164 M. Petraglia et al. through test ing. It ap pears that mod ern ag ri cul tural ac - Pub li ca tions, Hyderabad. tivi ties led to site destruc tion. ALLCHIN F. R. 1962. Painted pot tery from Patapadu, 4. Re touched tabu lar pieces con sist of flat and an - Andhra Pradesh. An tiq uity 36, 221–224. gular pieces of weath ered lime stone de rived from the BOIVIN N., KORISETTAR R., FULLER D. Q. 2005. bed rock found on the hillslope. The tab ular pieces of ten Further re search on the South ern Neo lithic and the display cor tex and pati na tion on their surfaces. Rel a - Ashmound Tradi tion: The Sanganakallu-Kupgal tively fresh flake scars along the perim e ter are often Archaeological Research Project, Interim Report. eas ily discernable. Jour nal of In ter dis ci plin ary Stud ies in His tory and 5. The limestone slabs were in a disturbed po si- Ar chae ol ogy 2, 63–92. tion, ly ing on the surface, as was also the case in Jwala- CAMMIADE L. A. 1927. Pre historic man in In dia and puram Local ity 12. Intact megalithic cists on the slopes the Kurnool bone caves: a ne glected field of in quiry. of ten con tain up right lime stone slabs, of ten placed in a Man in In dia 7, 105–111. par al lel or rect an gu lar pat tern. The Yaganti Lo cal ity 3 CHATTERJEE N., BHATTACHARJI, S. 2001. Pe - rockshelter contains an intact megalithic limestone cist. trol ogy, geo chem is try and tec tonic set tings of the 6. This is the same limestone bed rock unit that oc - mafic dikes and sills asso ci ated with the evo lution curs to the west, as docu mented at Jwalapuram Local ity of the Pro tero zoic Cuddapah Ba sin of south India. 20. Proceed ings of the In dian Acad emy of Sci ence 110, 433–453. Ac knowl edge ments CLARKSON C. 2002. An index of invasiveness for the measure ment of unifacial and bifacial retouch: a We thank the Ar chae o log i cal Sur vey of In dia for theoretical, ex per i men tal and ar chae o log i cal ver i fi- permis sion to carry out this research and the Ameri can ca tion. Jour nal of Ar chae o log i cal Sci ence 29, 65– In sti tute of In dian Stud ies (New Delhi) for ad minis tra - 75. tive sup port. Our re search has been supported by the DERANIYAGALA S. U. 1992. The Pre history of Sri Brit ish Acad emy, the Leakey Founda tion, Leverhulme Lanka: An Eco log i cal Per spec tive. De part ment of Trust, the Mc Don ald In sti tute for Ar chae o log i cal Re - Ar chae o log i cal Sur vey, Co lombo. search, and the Nat u ral En vi ron men tal Re search Coun - cil (NERC) (En vi ron men tal Fac tors in the Chro nol ogy FOOTE R. B. 1884a. Rough notes on Billa Surgam and of Human Evo lu tion and Dis persal programme), the other caves in the Kurnool Dis trict. Re cords of the NERC and Arts and Hu mani ties Re search Coun cil Ox - Geo log i cal Sur vey of In dia 17, 27–34. ford Ra dio car bon Ac cel er a tor Dat ing Ser vice, the Aus- FOOTE R. B. 1884b. Mr. H. B. Foote’s work at the tra lian Re search Coun cil, and Queens’ Col lege (Cam- Billa Surgam Caves. Re cords of the Geo log i cal Sur- bridge). Korisettar wishes to thank Karnatak Uni ver sity vey of India 17, 200–208. for sup port of this ar chae o log i cal pro ject. Shanti Pappu FOOTE R. B. 1885. Notes on the re sults of Mr. H. B. pro vided the GSI pub li ca tion that led us to the Jurreru Foote’s fur ther ex ca va tions in the Billa Surgam River Val ley. We thank members of the pro ject for Caves. Re cords of the Geo log i cal Sur vey of In dia their as sis tance dur ing field work, par tic u larly Ara- 18, 227–235. wazhi, Somashekharayya Hiremath Balageri, Kate FOOTE, R. B. 1916. The Foote Col lec tion of In dian Connell, Alison Crowther, Clair Har ris, Hannah James, Pre his toric and Protohistoric An tiq ui ties: Notes on Alex Mackay, Kalyan Malagyanavar, Abdul Munaff, their ages and dis tri bution . Gov ern ment Mu seum, Pragnya Prasanna, J. Varaprasada Rao and Ceri Ship- Ma dras. ton. Our field work could not have been con ducted FULLER D. Q. 2006. Ag ri cul tural Or i gins and Fron- with out the as sistance of vil lage sarpanches, Raga tiers in South Asia: A Work ing Syn the sis. Journal of Reddy (Betamcherla) and Konda Reddy (Pathapadu). World Prehis tory 20, 1–86. Ankala Reddy and nu merous other lo cal vil lag ers from FULLER D. Q., KORISETTAR R., VENKATASUB- Jwalapuram and Pathapadu pro vided us with in for ma- BAIAH P. C. 2001a. South ern Neo lithic Cul ti va tion tion and as sis tance for our ar chae o log i cal ex plo ra tions. Sys tems: A Re con struc tion based on Archaeobota- nical Ev i dence. South Asian Stud ies 17, 171–87. FULLER D. Q., VENKATASUBBAIAH P. C., KORI- REF ER ENCES SETTAR R. 2001b. The be gin ning of ag ri cul ture in ACHARYYA S. K., BASU P. K. 1993. Toba ash on the the Kunderu River Ba sin: ev i dence from ar chae o - In dian sub con ti nent and its im pli ca tions for cor re la - log i cal sur vey and archaeobotany. Puratattva 31, 1– tion of late Pleisto cene al luvium. Qua ter nary Re - 8. search 40, 10–19. FULLER D. Q., KORISETTAR R., VANKATASU- ALLCHIN F. R. 1960. Piklihal Ex ca va tions. Ar chae o - BBAIAH P. C., JONES M. K. 2004. Early plant do - log i cal Se ries No. 1. Andhra Pradesh Govern ment mes ti ca tions in south ern In dia: some pre lim i nary Investigations in the Kurnool Dis trict, South India 165

archaeobotanical results. Veg e ta tion His tory Archa- vium”. Qua ter nary Re search 41, 396–397. eobotany 13,115–129 MISHRA S., RAJAGURU S. N. 1996. Comment on FULLER D. Q., BOIVIN N., KORISETTAR R.. 2007. “New geo chem i cal ev i dence for the Youn gest Toba Dat ing the Neo lithic of South In dia: new ra dio met - tuff in In dia”. Qua ter nary Re search 46, 340–341. ric evi dence for key eco nomic, social and ritual MISRA V. N. 1989. Stone Age In dia: an eco log i cal trans for ma tions. An tiq uity 81, 755–778. per spec tive. Man and En vi ron ment 14, 17–64. GUPTA S., RAI S. S., PRAKASAM K. S., SRINA- MORRI SON K. D. 2007. Forag ers and for ager-trad ers GESH D., BANSAL B. K., PRIESTLEY K., GUAR in South Asian worlds: some thoughts from the last V. K. 2003. The na ture of the crust in south ern In - 10,000 years. In: M. D. Petraglia and B. Allchin dia: Im pli ca tions for Pre cam brian crustal evo lu tion. (eds) The Evo lu tion and His tory of Hu man Pop u la - Geo phys i cal Re search Let ters 30, 1419. tions in South Asia. Springer, Dordrecht, 321–339. ISAAC N. 1960. The Stone Age Cul tures of Kurnool. MURTY M. L. K. 1974. A late Pleis tocene cave site in Un pub lished PhD dis ser ta tion, Poona Uni ver sity. south India. Pro ceed ings of the Amer i can Philo - JAMES H. V. A., PETRAGLIA M. D. 2005. Modern soph i cal So ci ety 118, 196–230. hu man ori gins and the evo lu tion of be hav ior in the MURTY M. L. K. 1979. Re cent re search on the Up per later Pleis tocene record of South Asia. Cur rent An - Palaeo lithic phase in In dia. Jour nal of Field Archae - thro pol ogy 46, S3–S27. ol ogy 6, 301–320. JONES S. C. 2007. The Toba supervolcanic erup tion: MURTY M. L. K. 1981. Hunter-gath erer ecosys tems tephra-fall de posits in In dia and paleoanthropolo- and ar chae o log i cal pat terns of sub sis tence be hav - gical impli ca tions. In: M. D. Petraglia and B. iour on the south-east coast of In dia: an ethnogra- Allchin (eds) The Evo lu tion and His tory of Hu man phic model. World Ar chae ol ogy 13, 47–58. Pop u la tions in South Asia. Springer, Dordrecht, MURTY M. L. K. 1985. Ethnoarchaeology of the 173–200. Kurnool Cave ar eas, South In dia. World Ar chae ol - KORISETTAR R. 2007. To ward de vel oping a ba sin ogy 17, 192–205. model for Paleolithic set tle ment of the In dian sub - MURTY M. L. K. 1989. Pre-Iron Age agri cul tural set - con ti nent. In: M. D. Petraglia and B. Allchin (eds) tle ments in south In dia: an eco log i cal per spec tive. The Evo lu tion and His tory of Hu man Pop u la tions in Man and En vi ron ment 14, 66–81. South Asia. Springer, Dordrecht, 69–96. MURTI D. B. 2003. Mesoolithic cul tures. In: M. L. K. KRISHNA SASTRY, V. V. 2003. Mega lithic cul tures: Murty (ed.) Pre- and Protohistoric Andhra Pradesh the Iron Age. In: M. L. K. Murty (ed) Pre- and Up to 500 B.C. Andhra Pradesh History Congress, Protohistoric Andhra Pradesh up to 500 BC. Ori ent Hyderabad, 64–72. Longman, Delhi, 107–138. MURTY M. L. K., THIMMA REDDY K. 1975. The KUHN S. 1990. A geomet ric in dex of re duc tion for signif i cance of lithic finds in the cave ar eas of unifacial stone tools. Jour nal of Ar chae o log i cal Sci - Kurnool, In dia. Asian Per spec tives 18, 214–226. ence 17, 583–593. NAMBI K. S. V., MURTY M. L. K. 1983. An Up per LAKSHMI M. P., BABU H. V. R. 2002. Base ment Palaeo lithic fire place in Kurnool caves, south In dia. struc ture of the south west ern part of the Cuddapah Bulle tin of the Deccan College Re search Insti tute Ba sin from aero mag net ic anom a lies. Cur rent Sci - 42, 110–116. ence 82, 1378–1381. NASH D. T. 1993. Distin guishing stone ar ti facts from LAKSHMINARAYANA G., BHATTACHARJEE S., naturefacts created by rockfall processes. In: P. KUMAR A. 1999. Palaeocurrents and depositional Goldberg, D.T. Nash and M.D. Petraglia (eds) For- set ting in the Banganapalle For mation, Kurnool ma tion pro cesses in ar chae o log i cal con text. Prehis - Sub-ba sin, Cuddapah ba sin, Andhra Pradesh. Jour - tory Press, Madi son, 125–138. nal of the Geo log i cal So ci ety of In dia 53, 255–259. NEWBOLD T. J. 1844. Notes on the os se ous brec cia LUKACS J. R. 2007. In ter pret ing bi o log i cal di ver sity and de posits in the caves of Billa Soorgum. Jour nal in South Asian pre history: Early Holo cene pop ula - of the Asiatic Soci ety of Bengal 13, 610–615. tion af fin i ties and sub sis tence ad ap ta tions. In: M. D. PADDAYYA K. 1982. The Ach eul ian Cul ture of the Petraglia and B. Allchin (eds) The Evo lu tion and Hunsgi Val ley (Pen in su lar In dia)-A Set tle ment Sys - History of Hu man Pop ula tions in South Asia. tem Per spec tive. Deccan Col lege Post grad u ate and Springer, Dordrecht, 271–296. Re search In sti tute, Pune. LYDEKKER R. 1886. The fauna of the Karnul caves. PAPPU R. S., DEO S. G. 1994. Man-land re la tion ships Palaeontologia In dica Se ries X 4, 23–58. during Palaeo lithic times in the Kaladgi Basin, MISHRA S., RAJAGURU S. N. 1994. Comment on Karnataka. Deccan Col lege Post grad uate and Re - “Toba Ash on the In dian Sub con ti nent and Its Im pli - search In sti tute, Pune. ca tion for the Corre la tion of Late Pleis tocene Allu - PETRAGLIA M. D. 1995. Pur suing site for mation pro - 166 M. Petraglia et al.

cesses in India. In: S. Wadia, R. Korisettar and V. S. Cuddapah and Kurnool dis tricts. In dica 4, 75–94. Kale (eds) Qua ter nary En vi ron ments and Geoar- SARMA I. K. 1968. Patapadu re visited. A new painted chae ol ogy of In dia. Geo log i cal So ci ety of In dia, pot tery cul ture in South In dia. Puratattva, Bul le tin Bangalore, 446–465. of the In dian Ar chae o log i cal So ci ety 1, 68–84. PETRAGLIA M. D. 1998. The Lower Paleolithic of In - THIMMA REDDY K. 1968. Pre history of Cuddapah dia and its bear ing on the Asian re cord. In: M.D. Dis trict. Un pub lished PhD dis ser ta tion, Sagar Uni - Petraglia and R. Korisettar (eds) Early Hu man Be - ver sity. hav iour in Global Con text: The Rise and Di ver sity THIMMA REDDY K. 1977. Billasurgam: and Up per of the Lower Palaeolithic Record . Routledge Press, Palaeolithic cave site in South India. Asian Perspec - Lon don, 343–390. tives 20, 206–227. PETRAGLIA M. D., 2006. The In dian Acheulean in VENKATASUBBAIAH P. C. 1998. Re cent ar chae o- global per spec tive. In: N. Goren-Inbar and G. log i cal in ves ti ga tions in the Kurnool Dis trict, And- Sharon (eds.) Axe Age: Ach eul ian Tool-Mak ing hra Pradesh (a pre limi nary re port), Pro ceed ings of from Quarry to Dis card. Equi nox Pub lish ing, Lon - the Andhra Pradesh History Congress 22: 8–12. don, 389–414. VENKATASUBBAIAH P. C., PAWANKAR S. J., PETRAGLIA M. D., LAPORTA P., PADDAYYA K. JOGLEKAR P. P. 1992. Neolithic Fau nal Re mains 1998. The first Acheul ian quarry in In dia: stone tool from the Cen tral Pennar Basin, Cuddapah Dis trict, man u fac ture, biface mor phol ogy, and be hav iors. Andhra Pradesh. Man and En vi ron ment 17, 55–59. Jour nal of An thro po log i cal Re search 55, 39–70. WHEELER R. E. M. 1948. Brahmagiri and Chan- PETRAGLIA M., KORISETTAR R., BOIVIN N., dravalli. 1947: Mega lithic and other cul tures in the CLARK SON C., DITCHFIELD P., JONES S., Chitaldrug District, My sore State. An cient In dia 4, KOSHY J., LAHR M. M., OPPENHEIMER C., 180–310. PYLE D., ROB ERTS R., SCHWENNINGER J.-L., WIL LIAMS M. A. J., CLARKE M. F. 1995. Quater - ARNOLD L., WHITE K.. 2007. Middle Palaeo - nary ge ol ogy and pre his toric en vi ron ments in the lithic as sem blages from the In dian sub con ti nent be - Son and Belan Val leys, north-cen tral In dia. In: S. fore and af ter the Toba super-erup tion. Sci ence 317, Wadia, R. Korisettar and V.S. Kale (eds) Qua ter - 114–116. nary En vi ronments and Geoarchaeology of In dia. PRASAD K. N. 1996. Pleis to cene cave fauna from Geo log i cal So ci ety of In dia, Bangalore, 282–308. pen in su lar In dia. Jour nal of Caves and Karst Stud - ies 58, 30–34. PRASAD K. N., YADAGIRI P. 1986. Pleis to cene Cave Fauna, Kurnool District, Andhra Pradesh. Re - cords of the Geo log i cal Sur vey of In dia 115, 71–77. RAJU D. R. 1988. Stone Age Hunter-Gather ers: An ethnoarchaeology of the Cuddapah re gion, southeast In dia. Rav ish Pub lish ers, Pune. RAJU D. R., VENKATASUBBAIAH P. 2002. The ar - chae ol ogy of the up per Palaeo lithic phase in In dia. In: S. Settar and R. Korisettar (eds) In dian Ar chae - ol ogy in Ret ro spect: Pre his tory. In dian Coun cil of His tor i cal Re search, New Delhi, 85–109. RAO CH. VENUGOPAL, RAO C. V. N. K. 1992. Palaeontological stud ies of the cave fauna of Kur- nool District, Andhra Pradesh. Re cords of the Geo - log i cal Sur vey of In dia 135(5), 240–241. SAHA D. 2002. Multi stage de for mation in the Nalla- malai Fold Belt, Cuddapah Ba sin, South In dia – im - pli ca tions for Mesoproterozoic tectonism along the south east ern mar gin of In dia. Gond wana Re search 5, 701–719. SALI S. A. 1989. The Upper Palaeo lithic and Meso- lithic Cultures of Maharashtra. Deccan College Postgrad u ate and Re search In sti tute, Pune. SARMA I. K. 1967. Painted pot tery from Pusalapadu, Andhra Pradesh, and fur ther ex plo ra tions in the