TRADE, CONTACT, AND THE MOVEMENT OF

PEOPLES IN THE EASTER1 MEDITERRANEAN

STUDIES IN HONOUR OF J. BASIL HENNESSY

Edited by

Stephen Bourke and Jean-Paul Descreudres

MEDITERRANEAN ARCHAEOLOGY SUPPLEMENT 3

SYDNEY 1995

..

•

THE LAST HALF MILLION YEARS AT PELLA

Phillip C. Edwards and Phillip G. Macumber

INTRODUCTION

Ancient Pella, in the central Jordan Valley, holds an extraordinarily full record of human occupation, extending from about half a million years ago to the present day. Archaeological

Mediterranean Sea

Dead Sea

0 50 km

Figure 1. Sketch map of the southern Levant, showing the location of Pella in the Jordan Valley.

remains occur as numerous well-preserved s ites and innumerable scattered artefacts, distributed over several square kilometres surrounding the cold freshwater spring in Wadi Jirm al-Moz on the southern boundary of Tabaqat Fahl, and its hot-water companion, Hammamat Abu-Dhabli, two kilometres to the north in Wadi al-Hammeh (see fig. 2). This paper investigates the link between persistent water availability and human occupation through time.

For all the long time-span of occupation represented at Tabaqat Fahl , only the last decade has seen it placed in perspective, thanks to the foresight of Basil Hennessy's directorship of the Pella

project. In pursuance of the work reported here, the authors and their colleagues have been enabled to proceed due to Basil's constant enthusiasm and support.

Before moving on to a discussion of the Pleistocene remains at Pella (c .500,000-12,000 BP), we may accentuate the paper's theme by briefly listing the Holocene archaeol­ogical and historical periods represented at the site, from c.8,000 BP: Pottery Neolithic, 1

Chalcolithic,2 Early Bronze I/II,3 Early Bronze IV,4 Middle Bronze,5 Late Bronze,6 Iron

1 S. J. Bourke in: P. C. Edwards et al., ADAJ 34 (1 990) 61.

2 J. Hanbury-Tenison in: J. B. Hennessy et al., ADAJ 27 ( 1983) 325-31.

3 Bourke in: Edwards et al. art. cit. 61 - 2; P. M. Watson ibid. 76-80. 4 D. Petocz and L. Villiers, 'Wadi Hammeh survey', in: A.

MEDITARCH SUPPL. 3

McNicoll et al., ADAJ 28 (1984) 77-81; G. J. Wightman, 'The Early Bronze Age IV cemetery (Area XXXl)', in: T. F. Potts et al., ADAJ 32 (1988) 123-8.

5 T. F. Potts, 'The Middle Bronze Age (Area Ill)', in: Potts et al., ADAJ 29 ( 1985) 196-202. 6 Ibid.

2 P. C. Edwards and P. G. M~cumber

J 0

1R D A N

v A L L E y

r ,_; Wadi al - Hamrneh

LOWER PALAEOLITHIC MASHARIA Sl1ES

1 KM

N

I

Figure 2. The Tabaqat Fahl/Wadi al-Hammeh region showing Palaeolithic sites mentioned in the text.

Age,7 Hellenistic,8 Roman,9 Byzantine, 10 Umayyad, 11 Abbas id, 12 and Mamluk. 13 It is difficult to find another Levantine site with such a register. Yet, at Pella it comprises probably no more than one per cent of the time that hominids and humans have lived in the area.

7 A. McNicoll in: A. McNicoll, R. H. Smith, and B. Hennessy. Pella in Jordan I: an interim report on the joint University of Sydney and The College of Wooster excavations at Pella 1979-1981 (1982) 55-{)4; Bourke in: Edwards et al. art. cit. 62-3. 8 McNicoll art. cit. 65-75; J. C. Tidmarsh in: Edwards et al. art. cit. 71-{). 9 A. McNicoll in: McNicoll, Smith, & Hennessy op. cit. 77-101.

10 Ibid. 103-21; Watson toe. cit. (n. 3).

11 A. McNicoll in: McNicoll. Smith. & Hennessy op. cit. (n. 7) 123-4 I. 12 A. G. Walmsley in: A. W. McNicoll e t al., ADAJ 30 (1986) 182-95.

13 A.G. Walmsley and R. H. Smith in: A. W. McNicoll et al .. Pella in Jordan 2-The Second Interim Report of the Joint University of Sydney and College of Wooster Excavations at Pella 1982- 1985. Mediterranean Archaeology Suppl. 2 ( 1992) 183-98.

•

•

•

•

The Last Half Million Years at Pella 3

GEOMORPHOLOGY

THE TABAQAT FAHL

The Tabaqat Fahl (Plains of Fahl) region covers an area of about ten square kilometres in western Jordan, lying on the eastern edge of the Jordanian rift valley, about thirty kilometres south of Lake Tiberias (fig. 1). The Tabaqat is a large, buttress-like plateau which rises abruptly 125 m from the rift valley. Its eastern margin is the main eastern rift valley fault, where the landscape rises steeply towards Ajlun. The north and south boundaries of the Tabaqat Fahl are Wadi al-Hammeh and Wadi Jirm al-Moz respectively. Both bounding wadis are fed by large permanent springs-the Wadi Jirm al-Moz by the Pella Spring and Wadi al­Hammeh by the Hammamat Abu-Dhabli hot spring. These springs are the present-day representatives of spring systems which have been variously active over the last half million years, and strongly influenced human occupation and its later preservation in geological sequences.

The Tabaqat Fahl Formation forms the main geological unit underlying the Tabaqat. It consists of dense basal conglomerates, hard, dense limestone and thick tufaceous limestones. The conglomerate forms the basal unit in the sequence; it is largely composed of chert and limestone pebbles in a calcareous and siliceous matrix. The conglomerate caps the hills to the east of Tabaqat Fahl, and passes westward beneath the Tabaqat, where it is overlain and partly replaced by a tufaceous limestone which rapidly thickens towards the rift valley. At the rift valley edge, the underlying conglomerate is partially replaced by dense limestone. . The overall facies variation within the Tabaqat Fahl Formation represents a riftward change from ephemeral fluvial sedimentation (conglomerates) into a zone of massive spring deposition (tufaceous limestone) in a zone of regional groundwater outflow, occurring adjacent to a large lake (hard uniform limestone) within the rift valley.

This sequence has similarities to facies changes in the Late Pleistocene valley-fill deposits within Wadi al-Hammeh (dating between c.80,000 and 11,000 BP), which disconformably overlie the Tabaqat Fahl Formation (fig. 3). However the extent, both areally and temporally, of spring activity and tufa deposition required to create the bulk of the Tabaqat Fahl Formation is significantly greater than that which occurred in the later periods in Wadi al­Hammeh, and which are now represented by the Knob Limestone sequence (see below).

Lower Palaeolithic (Acheulean) sites are scattered throughout the middle Pleistocene Tabaqat Fahl Formation, which indicate an age for the top of the Formation of greater than 250,000 years BP. The lower age limit is unknown. However, given the thickness of the deposit, it may be in excess of half a million years.

THE WADIS

During Late Quaternary times riftward-flowing wadis in the central Jordan Valley incised deeply into pre-existing formations, and Wadi Jirm al-Moz and Wadi al-Hammeh both cut deep valleys into the Tabaqat Fahl Formation. These wadis were later backfilled with 50-60 m of alluvium, and these alluvial sequences were in tum incised at the end of the Pleistocene (c.11,000 BP) exposing the valley-fill deposits as remnants up to 50 m thick in section. 14

In Wadi Jirm al-Moz, Khirbet Fahl (the Pella tell ) has been built on Late Quaternary valley fill. 15 In Wadi al-Hammeh, a similar remnant valley fill now forms a high-level terrace,

14 P. G. Macumber and M. J. Head, ' Implications of the Wadi al-Hammeh sequence for the tenninal drying of Lake Lisan. Jordan '. PPP 84 (1991) 163-73.

15 P. G. Macumber, Geology of the Tabaqat Fahl area, northern Jordan (unpub. typescript, 1982).

4 P. C. Edwards and P. G. Macumber

which occupies most of the ancestral Wadi al-Hammeh. In the upper and central reaches of Wadi al-Hammeh the interfluvial terrace is bordered on the north by Wadi al-Hammeh and on the south by Wadi al-Rimar. On passing down-valley the terrace forms a narrow ridge which widens into an elongated flat plateau )qlown as the 'Plateau' (pl. 1: 1; fig. 4). This then passes via a small saddle into a small butte-the 'Knob'. Beyond the Knob the two wadis join and terrace remnants thereafter lie either side of Wadi al-Hammeh.

Cretaceous Umestone and Marl

TABAQAT FAHL

Tufaceous limestone

· } Taba.Qat Fahl Formation Conglomerate

WADI AL· HAMMEH SITES (Aged c. 12 · 80, 000 yrs BP)

~ WH27 WH26 WH34 WH 35 -41

MASHARIA SITES • Lower Palaeolithic

Natufian Kebaran Upper Palaeolithic Middle Palaeolithic

Figure 3. Schematic cross-section showing the relationship of the Late Pleistocene Wadi al-Hammeh sequences to the Middle Pleistocene Tabaqat Fahl Formation.

In 1980, large numbers of ¥tefacts were observed eroding from the valley fill in Wadi al­Hammeh. Over the next decade geological and archaeological surveys established the presence of numerous in situ Middle, Upper, and Bpi-palaeolithic sites occurring throughout the 50 m-thick valley fill, and extending for a distance of about two kilometres along Wadi al­Hammeh and its principal tributary, Wadi al-Rimar.

The archaeological sites are commonly associated with bands of the freshwater gastropod snail Melanopsis praemorsa, many of which were radiocarbon-dated. More than 35 dates have so far been obtained from both shell and charcoal.16 Taken together, the radiocarbon dates and their corresponding stone-tool industries show that the Late Pleistocene valley-fill deposit is the inland (continental) equivalent in age to the Jacustrine Lisan Formation deposited in Lake Lisan in the Jordan Valley from c.80,000 to 11,000 years ago. A date of 11,000 BP obtained from the uppermost sedimentary horizon in the wadi fill situated about one metre above the large Natufian site Wadi Hammeh (WH) 27-dated 12,000 BP­

indicated that this phase of sedimentation ceased at 11,000 BP. Incision immediately followed the drying of the lake at about 11,000 BP, leaving remnants of wadi-fill sediment now occurring as high-level terraces within the wadis.

16 Macumber & Head art. cit.

•

•

•

Wadi al-Hammeh

•

The Plateau

11,000

~~- 14,970 The Knob

27,000~ WH34 5 dates

®®®®®®®® (U.Pal.) ,0 28,550 to ~ 33,700

35,300 0 ® WH35 (M.Pal.)

Wadi Hammeh Conglomerates

Wadi al-Himar

® Archaeological sites 0 Discrete radiocarbon dates

~ Knob Limestone

~ Black Clay

D Wadi Hammeh Conglomerates

Vertical distance approx. 50m. Horizontal distance approx. 600m.

Figure 4. Diagrammatic section of the Plateau, showing the diachronous boundary between the Knob Limestone and the Wadi Hammeh Conglomerate (after Macumber & Head 1991 ).

~ ~

~ "' -~

<::::;:;

~ E:: cs· ;::s

~ $::)

;::; ~ -"ti ~ -~ en

6 P. C. Edwards and P. G. Macumber

GEOLOGY AND ENVIRONMENTAL RECONSTRUCTION

WADI AL-HAMMEH SEQUENCES

In the upper and middle reaches of Wadi al-Hammeh, the valley-fill sequence consists of pebble bands and conglomerates set in a red-clay matrix-the Wadi Hammeh Conglomerate. On passing downstream towards the valley near Masharia, a carbonate cement replaces the clay matrix of the Wadi Hammeh Conglomerate. The upper part of the conglomeratic sequence is gradually replaced by a tufaceous limestone (Knob Limestone) in a lateralfacies change.17

In some instances the two units are separated by a black clay. The boundary between the upper carbonate and the lower conglomeratic sequences is diachronous, spanning at least 20,000 years. It is older than 35,000 years downstream beyond the Knob, and about 15,000 years at its upstream limits close to the eastern end of the Plateau (fig. 3).

The Wadi Hammeh conglomerate was deposited as an ephemeral wadi-fill sequence, whereas the tufaceous Knob Limestone which appears nearer to the mouth of the wadi results from spring outflow in a broad zone of regional groundwater discharge occurring In the lower wadi, extending upstream for about a kilometre from the former Lake Lisan shoreline. The presence of discrete Melanopsis bands indicates permanent freshwater flow, emanating from within the spring zone. A similar situation occurs at the Pella springhead today. The general environmental picture is therefore one of an ephemeral wadi passing downstream towards a broad zone of permanent groundwater discharge, or freshwater spring outflow, in the lower parts of the wadi. The wadi then flowed into the ancestral Lisan Lake, with its shoreline lying against the edge of the Tabaqat at Masharia. 18 Lake Lisan was probably brackish in this region. 19

The diachronous lithological pattern represented by the Wadi al-Hammeh sequence is one stemming from a rising base level of sedimentation in response to the gradual filling of Lake Lisan. The rising lake level caused a concomitant rise in water tables, which in turn led to an upstream extension of the springhead and a gradual, time-transgressive onlap of the Wadi Hammeh Conglomerate sequence by the tufaceous Knob Limestone (fig. 3). At its thickest, on Wadi al-Hammeh's southern bank downstream of the Knob, the limestone is about thirty metres thick.

Sites in the calcareous silts (WH 31, 32, 51, 52) occupied a swampier, wetter area dominated by groundwater outflow. The physical setting would have been similar to that which exists around the Pella springhead today, where a number of small springheads coalesce to form the Wadi Jirm al-Moz. Sites occurring in the conglomerate/red-clay areas (WH 35-44, 49, 53) were established upstream of the main springhead. The black clay (containing the WH 26 site) which intermittently appears between the sequences, represents an intermediate environment. Its high organic content and the presence at times of a rich assemblage of Melanopsis molluscs suggests paludal conditions.

The presence of a large number of archaeological sites, spanning the latter part of the Middle Palaeolithic up to Natufian times, indicates that over this long period large springs provided a favourable occupational environment buffered against seasonal or longer-term aridity.

17 P. G . Macumber in: McNicoll et al. art. cit. (n. 4) 81--0; Macumber & Head art. cit.

18 P. G . Macumber in: McNicoll et al. art. cit. (n. 12) 156-7.

19 Z. B. Begin, A. Ehrlich, and Y. Nathan, 'Lake Lisan, the Pleistocene Precursor of the Dead Sea', Bull. Geol. Surv. lsr. 63 (1974) 1-30.

•

•

•

•

The Last Half Million Years at Pella 7

LOWER PALAEOLITHIC (MASHARIA SITES)

On the basis of the sedimentological similarity of the Tabaqat Fahl Formation to the Knob Limestones descdbed above, an examination was commenced of the former for archaeological sites in 1988. It immediately revealed five Acheulean sites, all of which contained bifaces, or handaxes.20 They have been termed the Masharia 1 to 5 sites, in order to separate them from the nearby but younger suite of Late Pleistocene Wadi al-Hammeh sites. The Masharia sites are named after the adjacent township of al-Masharia. The largest site, Masharia 1, occurs 70 to 80 m up a steep cliff face within a few hundred metres of the Masharia Post Office (fig. 2).

The five main sites are scattered widely across and back from the Tabaqat, being found close to its edges where erosion has exposed the sediments. Two of the sites (Masharia 1 and 3) are situated in the upper limestone member of the Tabaqat Fahl Formation towards the western end of the Tabaqat Fahl bluff, while Masharia 4 and 5 lie in the lower conglomeratic member in the central and eastern areas of the bluff. A fifth site, Masharia 2, also stratified in the lower conglomerate, was found in the cliff face of Tell Abu Rumayl, but has since largely been destroyed by agricultural activity.

The Masharia 1 site is located some twenty to thirty metres down a I 00 m-high cliff face, where large numbers of artefacts are visible, extending for almost 400 m along the section. At this site the artefacts occur in both solid, dense limestone and in earthy limestone, where they erode in large number to redeposit down the steeply sloping wadi bank. In a number of instances a dozen or more flakes are visible, packed together in solid rock, and in some cases, detached hard boulders contain strings of artefacts in an otherwise dense limestone matrix. The deposits are clearly in situ and contain material ranging from large handaxes to fine chipping debris. All the artefacts are sharp-edged and unabraded.

The Masharia 1 and 3 sites in the upper tufaceous limestone unit have each yielded numerous bifaces, commonly-but not always-with a white patina. They contain small cordiform bifaces, finely shaped, apparently by soft ~ammer percussion; numerous biface­thinning flakes, and at the Masharia 1 site elongate, pointed bifaces of a Micoquian character.21 These sites appear to be Late Acheulean in age, and by comparison with the Desert Wadi Acheulean sites from the Azraq Basin, may date to about 200,000 years BP.22

Two large handaxes occurred together at the Masharia 5 site in the lower conglomeratic member. Flaked cobbles and bifaces formed by hard hammer percussion were obtained from the Masharia 2 site, also in conglomerate, with the latter found at the base of the Tabaqat Fahl Formation. The close association in both instances suggests they had not moved far, if at all, from their original discard locations.

Stratigraphically, the Masharia 2 site, located at the base of the conglomerate, directly overlying Cretaceous sequences, is the lowest and oldest site in the Tabaqat Fahl Formation sequence. The two large handaxes from this site are significantly different in both patina and manufacture technique from those of the other sites. Given their position at the base of the sequence they may be considerably older than those from the Masharia 1 and 3 sites occurring in the upper limestone. Their age may exceed half a million years.

Archaeological and geological investigations, including radiometric and palaeomagnetic dating prograIJ1S, are planned for the Masharia sites in the near future. The fine stratigraphy of

20 P. G. Macumber, 'Geomorphology of Tabaqat Fahl ', in: A. G. Walmsley et al., ADAJ 37 (forthcoming); R. Y. S. Wright, 'The Acheulian of the Tabaqat Fahl Series', ibid. 21 R. V. S. Wright, pers. com.

22 L. Copeland and F. Hours (eds.), The Hammer on the Rock: Studies in the Early Palaeolithic of Azraq, Jordan (1989).

8 P. C. Edwards and P. G. Macumber

these sites promises to elucidate greatly the uncertainties over the relative time-placement of Lower Palaeolithic sites in the Jordan Valley.

MIDDLE PALAEOLITHIC

Interest has recently been rekindled in the L.evantine Middle Palaeolithic (c.150--40,000 BP23),

following the demonstration that both archaic Homo sapiens neanderthalensis and anatomically modem lfomo sapiens sapiens inhabited the region between hundred and forty thousand years ago. Successful application of thermoluminescence dating of burnt flint confirmed what some scholars had long suspected on the grounds of fauna} correlations, that the anatomically modem forms occurred earlier at Qafzeh Cave at c.92-90,000 BP,24 than the Neanderthals, who occupied Kebara Cave near Mount Carmel between c.60--48,000 BP.25

Moreover, both types of humans were associated with Levantine Mousterian flaked-stone industries, formerly thought to be correlated only with the Neanderthals.26

For the past sixty years most information on the Middle Palaeolithic has come from the deep deposits contained in the coastal limestone caves such as those at Mount Carmel.27 Less headway has been made in tackling the open-air agglomerations of Middle Palaeolithic material east of the Jordan River, that cover large areas of the Arabian Peninsula. This is because great complications and difficulties arise from establishing geological context, radiometric or stratigraphic control, and in modelling site definition for surface archaeological occurrences.

In Wadi al-Hammeh, Middle Palaeolithic chert artefacts are found scattered on the slopes and stream beds virtually continuously for a distance of two kilometres, from the Hammamat Abu Dhabli spring down to the mouth of Wadi al-Hammeh. These artefacts are continuously being shed from well stratified deposits, best exemplified in the archaeological horizons of the Plateau, where the lower twenty metres of sediments correspond to the latter part of the Middle Palaeolithic (c. l 00/80,000--40,000 BP). The uppermost Middle Palaeolithic site (WH 35) underlies a radiocarbon date of 35,000 BP (table 1, fig. 3). Thermqluminescence dating of one of the lowermost sites (WH 40) is currently being attempted in conjunction with the La . Trobe University Department of Geology dating laboratory. The unusual thickness of the Middle Palaeolithic deposits in Wadi al-Hammeh will provide an important comparative sequence· for less well developed wadi sequences and surface scatters in the arid Levant. The difficulty in determining the origin of the prodigious issue from the Plateau lies in the very profusion of artefacts. Accordingly, a series of small test spots were excavated into the westerly side of the Plateau deposits, at intervals of several metres (pl. 1: 1). Thirteen successive clay beds were tracked across the eroded westerly cliff face.

Artefacts were found in every one of the spots excavated between WH 41 and 35. Flaked stone artefact concentrations vary between hundreds and thousands per cubic metre, and the spots occasionally include pieces of burnt animal bone. Epicentres of material rarely appear, and conversely there is no place where considerable occupational debris does not occur.

The Wadi al-Hammeh Middle Palaeolithic assemblages are characterized by unipolar chert cores, many of which bear faceted striking platforms, used to produce blades, flakes, and triangular Levallois points.

23 B. Vandermeersch, 'The evolution of modern humans: recent evidence from Southwest Asia', in: P. Mellars and C. B. Stringer (eds.), The Human Revolution: Behavioural and Biological perspectives on the origins of modern humans ( 1989) 155-64.

24 H. Valladas et al .. Nature 33 1(1988)614-16.

25 H. Valladas· et al., Nature 330 (1987) 159-60.

26 J. J. Gowlett, Antiquity 61(1987)210-19.

27 D. A. E. Garrod and D. M. A. Bate, The Stone Age of Mount Carmel, 1 (1937).

..

..

"

•

The Last Half Million Years at Pella 9

Each spot has yielded sharp, unabraded artefacts, and small shatter flakes and chips, indicating that these materials were discarded at or near their ultimate find spots. WH 41 located at the bottom of the sequence in a fine silty colluvium, represents an extraordinary case of good preservation. A band of very sharp-edged flakes, blades, and knapping shatter was preserved in situ, extending little over a metre in section.

Many of the clay horizons also contain rolled artefacts with edges battered through water transport. These artefacts share technological features with the fresh lithics, but their condition is consistent with them having derived, through episodic rolling, from distances as short as 60 m from their eventual findspots,28 although many may have come from much greater distances and much earlier deposits. For any archaeological horizon, this mingling of human and natural agencies has resulted in an extensive palimpsest of material covering every comer of the valley floor.

Over the long periods during which each of the Plateau red clays was deposited, numerous occupational episodes have led to the formation of a mosaic of archaeological loci. In such cases we may find that the 'site' is in fact co-terminous with the extent of its enveloping deposit. The Wadi al-Hammeh MP sequences should be instrumental in elucidating the structure of well-preserved open-air Pleistocene sites that accumulate over long time-spans.

UPPER PALAEOLITHIC

WH 34 is a rich Upper Palaeolithic site stratified in a dark clay band sandwiched in the red pebbly clays of the Wadi Hammeh Conglomerate (pl. 1: 1). The site occupies a pivotal position in the local sequence, overlying the Middle Palaeolithic sites below it; and po·ssessing a bladelet technology characteristic of the much later Epipalaeolithic sites which overlie it. The site lies in a dark grey clay band, about two-thirds of the way up the exposed Plateau sequence, some eight metres below a Kebaran site (WH 26) discussed below. Radiocarbon dates sandwiching the site indicate an age of 30,000 BP

29 (table 1). The cultural deposit was sampled in successive spots across the exposed outcrop strip. The

lithics retrieved were sharp and included a high quantity of small-size fraction material, an entire series of bladelet core reduction products, and the highest lithic concentrations .of any of the sites discussed in the Wadi al-Hammeh, as well as traces of fauna. If the outcrop may

· be envisaged as the exposed strip of what was originally a circular or oval scatter, WH 34 equates to a site area of c.4,000 m2, by far the largest site of this period in the southern Levant.30 Using the same criteria of retrieval and measurement as for the other sites, WH 34 has clearly the highest lithic concentration (13,829/m3) of any of the sites investigated, being twice as dense as WH 31 and other later sites. This is somewhat surprising, considering the antiquity of the site and the scarcity of well-preserved Upper Palaeolithic sites. This does not include lithics recovered from flotation through a much smaller-gauge mesh (0.5 mm), further inflating the figure. Conditions of deposition are indicated as the main reason for the excellent preservation, since the site is embedded in a dark grey, silty clay, deposited under lower energy conditions than the red pebbly clays above and below it.

According to the prevailing view, assemblages of the Levantine Upper Palaeolithic can be divided into twofacies: Levantine Aurignacian and Ahmarian.31 The latter is characterized by having a numerical superiority of retouched blade and bladelet artefacts, and the former in

28 P. L. Harding et al. , ' The transport and abrasion of flin t handaxes in a gravel-bed river', in: G. de G. Sieveking and M. H. Newcomer (eds.), The Human Uses of Flint and Chert: Proceedings of the Fourth International Flint Symposium held at Brighto n Polytechnic 10-15 April 1983 ( 1987) 11 5-26.

29 Macumber & Head art. cit. (n. 14).

30 I. Gilead, 'The Upper Palaeolithic period in the Levant', JWP 5 (199 1) 105-54. '

31 I. Gilead in: J. Cauvin- P. Sanlaville (eds.). Pr~histoire du Levant (198 1) 331-42.

10 P. C. Edwards and P. G. Macumber

having a majority of heavier flake tools like burins and scrapers. Furthermore the distinction between the two facies has been attributed in part to the existence of two separate ethnic groups coexisting in the southern Levant at this time. 32

Further examination of Upper Palaeolithic assemblages may yet reveal that this supposed dichotomy is oversimplified and that in fact a continuous spectrum of core-reduction products and retouched implement types occur. The WH 34 cores consist almost entirely of small bladelet core remnants. Judging by the flake scars on these cores, blank shape was not accurately predetermined. Both wide flakes and thin, narrow bladelets were struck from the same cores. This condition is evident on the cores from the later Epipalaeolithic sites in Wadi al-Hammeh as well as the other suspected major Upper Palaeolithic site being investigated (WH 50).

For WH 34, most of the small number of retouched tools were burins made on flake blanks, along with a smaller number of scrapers and utilized chunks registered as scaled and battered pieces. On the other hand WH 50 (for which no radiocarbon dates have yet been obtained), contains a component of bilaterally retouched and pointed bladelets. Both sites have a similar range of core remnants. The possibility of carbon-dating all the Wadi al­Hammeh Upper Palaeolithic sites will produce significant new data for addressing the issue of Levantine Upper Palaeolithic taxonomy.

EARLY EPIPALAEOLITHIC (KEBARAN)

During the Late Glacial Maximum (c.18-16,000 years ago), ice sheets in the high latitudes reached their maximum extent.33 Coincidentally the Middle East underwent. some of the coldest and· driest conditions for which we have evidence in the region.34

Three Early Epipalaeolithic (Kebaran) sites have been excavated in Wadi al-Hammeh that date to this cold, arid phase. The first of these, WH 26, is stratified in a silty black clay near the· top of the Plateau (89 m below sea level), some eight metres above the above-mentioned WH 34. A radiocarbon date of c.19,500 BP was obtained for WH 26 from wood charcoals. These fragments are the remains of a hearth, around which were clustered rich quantities of lithics and animal bone fragments extending about 25 m2 in area.35

The charcoals indicate that the inhabitants of the site gathered and burned wood from a variety of trees and shrubs. The fragments come from wild oak (Quercus sp.), almond (Amygdalus sp.), and pistachio (Pistacia sp.) trees, as well as from hawthorn (Crategus sp.), buckthorn (Rhamnus sp.), and hackberry (Ce/tis sp.) shrubs.36 Today, none of these plants grow in Wadi al-Hammeh. Rather, the tree line begins some two kilometres up-valley to the east, about I 00 m above sea level. Even at the most arid period of the Pleistocene then, the well-watered Wadi al-Hammeh provided a buffer to dry conditions. Moreover, oxygen isotope analyses indicate th~t mean temperatures in the region were depressed by about 4° at

32 Ibid.

33 C. Gamble and 0 . Soffer, 'Pleistocene polyphony: the dive rs ity of human adaptations a t the Last Glacial Maximum', in: C. Gamble and 0. Soffer (eds.), The World at 18,000 BP, II: Low Latitudes (1990) 97- 118. 34 S. Bottema and W. Van Zeist, ' Palynological evidence for the climatic history of the Near East, 50,000-<i,OOO BP', in: Cauvin & Sanlaville (eds.) op. cit. 11 1- 32; U. Baruch and S. Bottema, 'Palynological evidence for climatic changes in the Levant ca. 17,000-9,000 B.P.', in : 0. Bar-Yosef and F. R.

Valla (eds.), The Natufian Culture in the Levant (1991) 11-20; M. Weinstein-Evron, 'Palynological history of the Last Pleni glacial in the Levant ', in: Colloque sur Jes industries A pointes foliac~es du Paleolithique Sup~rieur

euro¢en, Cracovie (1990).

35 P. C. Edwards, 'Kebaran occupation at the Last Glacial Maximum in Wadi al-Hammeh, Jordan Valley', in: Gamble & Soffer (eds.) op. c it. 97- 118.

· 36 G. Willcox, 'Preliminary report on Plant Remains from Pe lla', in: McNicoll et al. op. cit. (n. 13) 253-6.

•

•

The Last Half Million Years at Pella 11

this time.37 The wadi's low altitude and the presence of a permanent (and hot!) water source ameliorated the cold, arid conditions. ·

During this period the inhabitants of Wadi al-Hammeh hunted a broad spectrum of animal prey,38 including wild pig, gazelle, ovicaprids, fox, wild cat, hare, tortoise, and partridge.

In a technological sense the flaked-stone industry of the Kebaran sites shows strong continuity from the Upper Palaeolithic site WH 34. Alluvial chert cobbles were procured from the wadi to manufacture blade and bladelet cores in the Kebaran sites WH 26, 52, and 31, dated respectively c.19,500, 19,480, and 16,700 BP. Generally, this was accomplished by removing the top of a cobble to form a simple striking platform, then spalling around a section of the core face below the platform. Cores were rarely heavily reduced, reflecting the abundance of readily available local materials. Like the Upper Palaeolithic practice, blank shape was not closely controlled, so that many wider flakes were detached along with blades and bladelets.

Considerable variability is evident amongst the retouched tool components from the three Kebaran sites. WH 26 (c.19,500 BP) mostly contains non-geometric microliths, (c.80 per cent of the retouched artefact total), and of these, the most common type is the obliquely-truncated backed bladelet. The contemporaneous Kebaran site WH 52 is situated a kilometre down­valley, embedded in the fine, grey sediments of the Knob Limestone, and when occupied it would have overlooked the Lake Lisan shore. This site contains burins and scrapers, but almost entirely lacks a microlithic component. WH 26 is a much richer site than WH 52, and further analyses should throw light on whether functional differences (that is, varying uses to which tools were put), can explain the divergence between the· two sites, or whether assemblage size has contributed an effect.

WH 31 is stratified above WH 52 at the mouth of Wadi al-Hammeh, and WH 31 is in turn overlain by a date of c.16,700 BP. Like WH 52 and the other sites excavated in the Knob Limestone, WH 31 lacks any substantial fauna! component, making direct comparisons of subsistence behaviour with WH 26 difficult. The contrast most probably results from the mineral composition of the Knob Limestone.

WH 31 shares similarities in lithic technology to the other Kebaran sites, but introduces novel typological features which seem to be the result of the site's younger age. The majority of microliths are backed bladelet fragments, some with straight truncations at one end. In two cases bladelets are bi-truncated to form elongated rectangular forms, perhaps presaging the shift to geometric microlith production, widespread in the Levant after c.16,000 BP.39

In addition to showing considerable intra-site typological variation, tl)e Early Epipalaeolithic Wadi al-Hammeh sites add further data to the growing awareness of inter­regional variation in the Early Epipalaeolithic, a trend that is already clear for the Late Epipalaeolithic. The typology described for the Wadi al-Hammeh Early Epipalaeolithic sites is quite distinct from contemporaneous sites elsewhere, such as Uwaynid 14 and Jilat 6 in the

37 Z. Reiss et al., 'Paleoceanography of the Gulf of Aqaba during the last 150,000 years', Palaeoecology of Africa and the surrounding islands 16 (1984) 55--{)5. 38 P. C. Edwards, 'Revising the Broad Spectrum Revolution and its role in the origins of Southwest Asian food production', Antiquity 63 (1989) 225-56; K. V. Flannery. 'Origins and ecological effects of early domestication in Iran and the Near East', in: P. J. Ucko and G. W. Dimbleby (eds.),

The Domestication and Exploitation of Plants and Animals (1969) 73-100.

39 A. N. Garrard and B. F. Byrd, 'New dimensions to the Epipalaeolithic of the Wadi Jilat in Central Jordan', Palcforient 17/2 (in press); 0. Bar-Yosef and A. Belfer­Cohen, 'The origins of Sedentism and Farming Communities in the Levant'. JWP 3/4 (1989) 447-98; Gilead art. cit. (n. 31).

12 P. C. Edwards and P. G. Macumber

Azraq Basin,40 the northern Palestinian cluster of sites,41 Urkan-e-Rubb on the western side of the Jordan Valley,42 and Ohalo II on Lake Tiberias .43

LATE EPIPALAEOLITHIC (NATUFIAN)

The Early Natufian site WH 27 (c.12,000 BP) which caps the long Pleistocene Plateau sequence is the most complex of the Wadi al-Hammeh sites. Its form, in comparison to the long succession of sites stratified beneath it, parallels in microcosm the important changes which occurred in human settlement systems in the Levant at the end of the Pleistocene. At this time a shift occurred from a highly mobile settlement system, characterized by short-term sites, to a less mobile system in which more energy was expended on the construction and maintenance of residential sites.44

Broad-area excavations at WH 27 have recovered the remains of two large curvilinear structures. The first (Structure 1) is a horseshoe-shaped structure, delineated by a terraced wall stub of limestone blocks. Its south-western margin lies open and is demarcated by several stone-ringed post holes. Excavations through Structure 1 in Plot XX F show three successive, superimposed constructional phases. In each the similar placement of architectural features indicates continuity of use at the site through several rebuilds.45 At the base of this sequence a young adult male was buried in a pit cut into the natural travertine. A second, apparently multiple, burial was salva~ed from beneath the architectural remains along the eroding westerly cliff section of the site.4

The second dwelling (Structure 2) is built in a similar manner to the first, albeit with a completely enclosed outer perimeter, except where erosion has cut the structure to the west. The perimeter wall traces out a large oval measuring 13 x 9 m. Two enigmatic concentric wall sections lay within this, and one of these terminated with three large slabs decorated with incised concentric decorations. Several other semi-naturalistic and geometric incisions were found on plaques, shaft straighteners, and small fragments of limestone.

Stone features such as circles, post holes, and hearths occur within and without the two large structures. Huge quantities of rubbish were deposited across the site, most notably within the dwellings. Besides large numbers of lithics and faunal fragments, several discrete artefact groups were found cached against the inside of the perimeter walls in both structures. These include basalt plant milling gear (pl. 1: 2); a grouped arrangement including a sickle, lunates, polished stones, a core, and gazelle podials; further piles of gazelle podial bones, and caches of large flaked stone tranchet axes. The variety of implement and ornamental classes testifies to the varied residential and domestic activities carried out on the site.

The complex mosaic of microclimatic habitats bordering the Lisan shore is reflected in the taxonomically diverse fauna recovered from WH 27. Its residents preyed on virtually every creature that walked, crawled, or flew, including gazelle, wild aurochs, pig, equid, red deer, roe deer, fallow deer, ovicaprines, fox , hare, tortoise, freshwater crab, partridge, stork, coot, owl, gull, and duck.47

40 B. F. Byrd in: 0 . Aurenche, M.-C. Cauvin, and P. Sanlaville (eds.), Pr~histoire du Levant: processus des changements culturels ( 1991) 257--04.

4 1 0. Bar-Yosef, 'The Epi-Palaeolithic complexes in the Southern Levant', in: Cauvin & Sanlaville (eds.) op. cit. (n. 31) 389-408.

42 E. Hovers et al .• Mitekufat Haeven 21 ( 1988) 20-48.

43 D. Nadel, Mitekufat Haeven 23 (1990) 48-59.

44 B. F. Byrd, 'The Natufian: settlement variability and

economic adaptations in the Levant at the end of the Pleistocene', JW P 3/2 ( 1989) 159- 97.

45 P. C. Edwards, 'Natufian settlement in Wadi al-Hammeh". Pal~orient 14/2 (1988) 309- 15.

46 S. J. Bourke in: A. N. Garrard and H. G. Gebel (eds.), The Prehistory of Jordan: The State of Research in 1986 ( 1988) 558- 9.

47 P. C. Edwards et al. in: Garrard & Gebel (eds.) op. cit. 547-9.

..

•

The Last Half Million Years at Pella 13

Extensive flotation of sediments from WH 2748 was rewarded by the recovery of carbonized botanical fragments, rare for this period. These include wild barley, lilies, legumes, and parasitical plants such a8 dodder (Cuscuta sp.). Several more species remain to be identified, and work in this important area is proceeding.49

SITE M.B.S.L. PERIOD DATE (BP)

Wadi Hammeh 27 83.5 Late E. P. (Early Natufian) 11,920 ± 150, (0xA393) 11,950 ± 160 (OxA507) 12,200 ± 160 (OxA394)

Wadi Hammeh 31 153.7 Early E. P. (Kebaran) > 16,740 ± 220 (ANU4654) Wadi Hammeh 26 89.2 Early E. P. (Kebaran) 19,500 ± 600 (SUA210l) Wadi Hammeh 52 139.5 Early E P. (Kebaran) 19,480 ± 500 (ANU4653) Wadi Hammeh 33 89.8 Early E. P. (Kebaran) 19,500 ± 600 (SUA2101) Wadi Hammeh 5 l 156.1 Early E. P./U. P? Wadi Hammeh 32 141.6 U.P. > 19,480 ± 500 (ANU4653) Wadi Hammeh 34 96.1 U.P. 28,550 ± 550 (ANU5826)

28,940 ± 900 (ANU4657) 33,700 ± 700 (ANU700)

Wadi Hammeh 50 105.9 U. P.? Wadi Hammeh 42 98.0 U. P./M. P? Wadi Hammeh 43 99.7 U. P./M. P? Wadi Hammeh 44 101.3 U. P./M. P? Wadi Hammeh 35 102.3 M.P. >35,300 + 1320 /-1130 (ANU583 l) Wadi Hammeh 36 105.0 M.P. Wadi Hammeh 37 108.2 M.P. Wadi Hammeh 38 113.3 M. P. Wadi Hammeh 39 116.1 M.P. Wadi Hammeh 40 121.4 M.P. Wadi Hammeh 41 122.4 M.P. Wadi Hammeh 49 115.9 M.P. Wadi Hammeh 53 ll3.8 M.P. Wadi Hammeh 54 117. l M.P.

E.P. ~ipalaeolithic U.P. M:per Palaeolithic M.P. ddle Palaeolithic M.b.s.I. Met~es below sea level

Table 1. Excavated Pleistocene sites in Wadi al-Hammeh .

CONCLUSION

The repeated cycles of human occupation in the Pella region are measured in the ubiquitous sites and artefacts within its long sequences of waterlain sediments. The depositional environments provided unusually favourable conditions for preservation. They also reflect the persistence of fluviatile, marsh, spring, and lakeside environments through much of the Pleistocene that continually attracted people to the area.

Consequently, the Pleistocene record of the Tabaqat Fahl/Wadi a l-Hammeh region includes Lower Palaeolithic sites dating from c.500,000 to c.250,000 BP, Middle Palaeolithic

48 S. M. Colledge in: Garrard & Gebel (eds.) op. cit. 54i-56. in Epipalaeolithic sites in the Near East', in: Bar-Yosef & 49 S. M. Colledge. ' Investigations of plant remains preserved Valla (eds.) op. cit. (n. 34) 391 - 8.

14 P. C. Edwards and P. G. Macumber

sites from c.100,000 to 40,000 BP, Upper Palaeolithic sites dating c.30,000 BP, Early Epipalaeolithic sites dating c.19,500-16,750 BP, and a Late Epipalaeolithic (Natufian) site dating to 12,000 BP.

The latter site marks an important crossroads for the area, geologically and historically. Towards the end of the Pleistocene, and after the abandonment of WH 27, the adjacent Pleistocene Lake Lisan dried up at c.11,000 BP.so This event resulted in a lowering of the water table. Subsequently an erosional cycle was set in motion in the riftside wadis , producing deep incision into the former Pleistocene valleys.

In this post-Natufian, early Holocene period, a new form of settlement became established, by settled hunters who cultivated domestic plants.SI

The craggy outcrops of Wadi al-Hammeh, perched far above the water table, were now unsuitable for the positioning of agricultural villages and cultivated fields. Settlement thus shifted southwards two kilometres, to the springhead of Wadi Jirm al-Moz. There it has remained, more or less continuously, to the present day in the form of the Khirbet Fahl tell and its satellites.

50 Macumber & Head an. c it. (n. 14).

51 0. Bar-Yosef and M. Kislev, 'Early farming communities in the Jordan Valley', in: D. R. Harris and G. C. Hillman

(eds.), Foraging and Farming: th e Evol uti on of Plant Exploitation (1989) 632-42.

. . . p

P. C. Edwards and P. G. Macumber PLATE 1

•

1. Palaeolithic sites stratified in the Plateau; Pella, Jordan

2. cache of basalt mortars. mullers, and a pestle from Phase I of the Natufian site Wadi Hammeh 27.