Durham Research Online

Deposited in DRO:

01 September 2009

Version of attached �le:

Published Version

Peer-review status of attached �le:

Peer-reviewed

Citation for published item:

Wilkinson, T. J. and Edens, C. and Barratt, G. (2001) 'Hammat al-Qa : an early town in southern Arabia.',Proceedings of the Seminar for Arabian Studies., 31 . pp. 249-259.

Further information on publisher's website:

http://www.arabianseminar.org.uk/psas31.html

Publisher's copyright statement:

Additional information:

Use policy

The full-text may be used and/or reproduced, and given to third parties in any format or medium, without prior permission or charge, forpersonal research or study, educational, or not-for-pro�t purposes provided that:

• a full bibliographic reference is made to the original source

• a link is made to the metadata record in DRO

• the full-text is not changed in any way

The full-text must not be sold in any format or medium without the formal permission of the copyright holders.

Please consult the full DRO policy for further details.

Durham University Library, Stockton Road, Durham DH1 3LY, United KingdomTel : +44 (0)191 334 3042 | Fax : +44 (0)191 334 2971

https://dro.dur.ac.uk

l:Iammat al-Qic: an early town in Southern Arabia

T. J. Wilkinson, C. Edens & G. Ba"att

By autumn 1999, a team from the Oriental Institute,UniveBity ofChicago bad conducted five seasons ofsurvey and soundings in the Yemen highlandsaround Dbamir. This work has provided a basicarchaeological and environmental framework for theregion as welJ as some general information on thenature of the landscape within which thearchaeological sites were found. In 1999 we thenstarted more specialized studies, initially with thesite of ijammat al-QI<, the ground plan of whichwas recorded stone-by-stone to reveal a complexand dense pattern of buildings over an enclosed areaof 3 ba, and a total site area of 5 ha. Chronologicalcontrol is provided by radiocarbon assay ofexcavated cbarcoaJs and a local ceramic sequence.We also supply data on the agrarian economy usingoff-site features from relict terraced fields andthreshing floors and then estimate the cultivated arearequired by the inhabitants in order to attempt apreliminary reconstruction of the Bronze Age landuse. I

latroduc:tioaAlthough there is no evidence for towns in Arabiabefore the third millennium BC, by the end of thatmillennium large settlements were evident at 11J1ltIsland in Saudi Arabia (potts 1990, i: ~7. lOS­109), Qa1<at aI-Bahrain (H8jlund &; Andersen 1994;1997). Saar (also OIl Bahrain: Cnwford, Killiet clMoon 1997), Ten Abraq (potts 1991) and Hili in theUAE (Cleuziou 1982), and Bit in Oman (Frifelt1976) (Fig. I).

In Yemen, pre-Iron Age settlement was firstdocumented in 1981 (de Maigret 1984). in theKhawlln district of the eastern highlands, where anItalian team investigated numerous small thirdmillennium settlements (de Maigret 1990). Similarsmall sites sharing the same architectural andceramic tradition appear on the desert fringes ofinterior Yemen (e.g. Blakely. Vitaliano &; Brinkman1996; Cleuziou. Inizan &; Marcolongo 1992; Breton&; Darles 1994). On the Red Sea Tibama coast.unrelated pottery appears in large shell middens ofthe second (and perhaps third) millennium (Zarins &;

Zabrani 1985; Zarins &; aI-Badr 1986; Tosi 1986),and recent investigations at Sabir near Aden haverevealed a massive town of the late secondmillennium and early first millennium BC (Vogt clScdov 1998).

In those parts of Yemen where rainfall in excessof 300 rom per annum falls on intermontaneplateaux between 2000 and 3000 m above sea level.archaeological surveys suggest the existence ofnumerous small Bronze Age towns. Unlike thesettlements of the Gulf andso~ Arabia. thoseof Yemen are OIl hilltops and are distinguished bythe clarity of their building layouts which consist ofalignments of collapecd and heavily weatheredwalls. A key factor for aettlemeot is that. as a resultof their location in the zone of sununer monsoooa1rainfall. these highlands are the best watered part ofthe Arabian peninsula. The potential for significantcrop productioD is high as long as there is sufficientlOil for cultivation. a deficit that is usually Ilddressedby the coostruetion of terraced fields.

249

T.J. WILKINSON, C. EDENS & G. BARRATT

Nl'

o 500kmI I

FIGURE 1. Location ofkey sites mentioned in the text. Inset shows the location ofthe Khawl4n sites and someofthose ofthe Dhamlir Project. First published in Antiquity 74.2000.

Dhamer Bronze Age attes

16

H

12

10

!J

8

8

..2

0HnlIglr ........ Med.

OIl 8lBeIed

FIGURE 1. Bar chart showing relative areas (in ha) ofthe major Bronze Age sites in the Dham4r area.

250

I;IAMMAT AL-QAc: AN EARLY TOWN IN SOlITHERN ARABIA

To date the Oriental Institute Project for theArchaeology of Yemeni Terraced Agriculture hasrecorded 56 Bronze Age sites, several of whichexceeded 3 ha in area. Noteworthy sites include:Kharraib (OS 228), Medinet al-Balad (OS 187), al­Sibal (OS 66), Hawagir (OS 293), al-MosaynCa (OS268), and I;Iammat al-QIc (OS 101) (Figs. I and 2).The local Bronze Age pottery belongs to the sametradition found in the drier eastern highlands andinto the interior desert fringe (Khawlln, Jawf, WldIal-JQbah, Widi Markha) although the Dhamirpottery also displays some distinctive characteristics(Edens 1999).

The Dhamir sites occur mainly in the semi-aridpart of the highlands where today mean annualrainfall is between 250 and 350 mm. In the north­east part of the survey area where rainfall is less,scattered Bronze Age villages are evident, whereasto the south-west where rainfall is higher BronzeAge sites are less visible. This is because in thewetter highlands where terraced fields extend up tothe hill summits, sites have been transformed as a

ncURE 3. Site ofl;Iammat al-~ and adjacent off­site areas on slopes. Contours at 5 m intervals; Th=Threshingfloor. First published in Antiquity 2000.

- ...'--=-.0.,_-,:-. -­1"""".1

result of taphonomic processes such as the removalof stones for the construction of the walls of fields.

For the most part, the Bronze Age sites areshallow and lack later occupation. Where adversetaphonomic effects have not 0CC'UJTed, the stonearchitecture of many sites is readily apparent andtherefore large portions of these sites can bemapped. l:Iammat a1-Q1c provides an exceptionallywell-preserved example of this kind.

aammat al-Qic: The sitel:Iammat a1-Q1c was discovered by analysis of aerialphotographs prior to the 1995 field season. Thesurvey team visited J:Iammat al-QIc in 1995, anddated the site to the third millennium BC bycomparison of surface pottery with that of theKhawlln sites (Wilkinson, Edens & Gibson 1997).The site sits upon a gently inclined but otherwiseflat-topped hill the summit of which is 50 m abovethe level of a narrow plain (Fig. 3). This surface isthe remnant of a basalt flow that overlies. in turn. •conglomerate of alluvial cobbles and a basic igneousformation. The basalt cap provided building materialfor the town, and the conglomerate felsite offeredcobbles for a chipped stone industry.

251

TJ. WILKINSON, C. EDENS cI: G. BARRATT

During the 1996 season preliminary mappingidentified the basic architectural components of thesite, and five soundings acquired pottery andradiocarbon samples in association with the townwall and several buildings. In addition, apreliminary site map identified the basiccomponents of the site. Two additional soundingsand detailed mapping of the entire site and itssurrounding area in 1999 completed the fieldinvestigation to date. These investigations date thesite to the late third and first balf of the secondmillennium BC (2320 to 1410 caIBC; Wilkinson &Edens 1999, table 2), that is a little later than theKhawlln sites (de Maigret 1990) and al-Sibal in theDhamlr area, and somewhat earlier than Kharraiband Hawagir in the DbamIr area (Wilkinson &Edens 1999, Edens 1999). A sounding excavatednext to the south-west town wall revealed twophases of construction separated by a spread ofrubble. The rubble contained a mix of Bronze Ageand later pottery, suggesting that the wall was rebuiltearly in the first millennium AD or even later.Occasional Iron Age and Himyarite sberds on thesurface, a single first millennium BC radiocarbnndate, and the apparent remodelling of somearchitecture also indicate that there was some latere-use of the site. The excavations provided a faunalassemblage of sheep/goat (90010) and cattle (10-10),but recovered no arcbaeobotanical remains besidesrare wood charcoal.

TOWD FabricThe site exhibits several features that arecharacteristic of an urban layout rather than avillage, namely: a surrounding wall with gates, adense layout of buildings and a possible subdivisioninto quarters (Fig. 4). An outer defensive wall, 1.8 mwide, is visible almost continuously around thesouth-eastem half of the hill but only intermittentlyaround the northwestern perimeter. A 1.4 m thickcross-wall runs from nortb-east to south-west acrossthe hill. cutting the site into two unequal parts (Fig.4). The dense architecture of the main site covers thelarger south-east area of 3.0 ha, while a thinnerscatter of buildings characterizes the remaining 1.9ha of the hill (the north-west site). Ash middens thatrun up against the interior face of the perimeter wallon the north-east side of the main site provided two

252

radiocarbon dates in the range 141()-1960 calBC(trench 3; Fig. 4: TI), firmly relating this wall withthe Bronze Age town. Although the cross-wall. withits different construction style, has not yet beendirectly associated with the town, the adjacentBronze Age structures were not robbed of theirstones for construction of the wall, stronglysuggesting that it too belongs to the Bronze AgesettIemenL

Two gates offer access through the north-eastperimeter wall. Both are framed by large ortbostats,the more northerly gate I being 3.2 m across, andthe more southerly gate 2 1.5 m wide (Fig. 4: G Iand G2). A less obvious gate (G3) punctuates thesouth-west perimeter wall where a stairway ofseveral risers leads up to a gap in the wall. Since thispart of the wall was rebuilt long after the BronzeAge, this gate does not certainly belong with theBronze Age town.

Despite first impressions of disorder, the town iswell-structured at several levels. Built-up zonesalternate with streets and open spaces to impose aperceptible order to the site as a whole. A wallcurves westward from the perimeter wall to thecross-wall, closing off the northern comer of thetown, the highest point of the butte (Fig. 4: W). Astreet runs along the outside face of this wall fromgate I, and a topographic indentation half way alongthe wall may mark the entrance to this uppercompound from the street. A strip of buildings onlow terraces extends along the perimeter wall south­east of gate I, forming a distinct zone about 80 mlong and 25 m deep. In contrast. the south-westperimeter wall borders a large zone of open space inthe north-west quadrant of the site, and a scatter ofbuildings farther south.

Between these two sides of the town, sevenelongated blocks of housing~ m by 25-30 m,alternate with streets and strips of open space, allbeing oriented with the site slope. The blocks eachcontain from two to seven buildings; the Itrip south­east of gate I encompasses at least nine buildingsand the remaining zones C8Ch bold another four orfive structures. Open strips. I()-IS m wide anddevoid of architectural rubble, IepInIe the n0rth­western blocks while NE-SW Itreeb run betweenthe remaining blocks; NB-SW streets delimit theblocks from architecture aloog the nortb-eut and

J:IAMMAT AL-QA<: AN EARLY TOWN IN SOUTHERN ARABIA

N

t

o

nCURE 4. Overall map ofthe site showing all extant walls and gates G1-G4. Note that only those parts ofthetoWli wall tJtat were mapped in 1999 are indicated. W is the perimeter wall of the 1Iorthern qauuter of thesettkmDtl; T3 nfers 10 Tr6IClI J e:xctJVQJed in 1996; G refers to Gales. Fint paJHislted in Antiquity 74.1000.

south-west perimeter wall. The streets generallyhave gravel and fine rubble surfaces and boundariesof low curbing or retaining walls.

Individual houses mainly consist of free-standingrectilinear struetun=s separated ooe from another byirregular intervals; nearly all are orientedDO~IOUIb-west in KCOrdancc with the site slope (Fig. 5).Some buildings are linked by retaining walls orrough linea of stones that run perpendicular to theslope. 1'beIe linbaes. together with the arrangementof buildinp within blocks, hint at the existence ofsupra-residential social units. One block in thecentral main site offers a view of this organization.Here 6-7 houses are grouped around a central openIp8Ce (Fia. S: B). with a lane entering the opensplICe &om the street to the north (A), and threeretainina waIIJ (a *pped taT1lce?) connecting thetwo boutea at the IOUIb-west end of the block (C).

Outside the croa-wall to the nortb-wcst. thedensity of I1IuctUra iI much less. Between 16 and22 buildings are sc:aaercd through this nortb-west

site, with relatively dense clusters occupying thewestern point of the hill, the eastern comer near thecross wall, and the central area (Fig. 4). The layoutof individual houses is far less obvious than in themain site, and there is less wall collapse and rubble.This absence of stones and rubble suggests that thecomponent stones have been robbed out, perhaps foruse in other parts of the site.

Unambiguously complete plans are available for16 houses in the main site and another seven for thenorth-west site. Almost all the complete houses aresimple rectangular or 'L'-shaped structures. Thebasic house is rectangular, with the interior spaceeither being a single room or su1H1ividcd intoseveral rooms by cross-walls. Buildings aretypically naITOW (2-5 m), but may be up to 30 m inlength. Outer walls are of basalt blocks laid incourses or slabs set on edge; they are usually onecourse wide resting upon a foundation of smallerstones.

253

T.J. WILKINSON, C. EDENS & G. BARRA17

•t

o 10.

nGURE S. Detail showing buildings in the south-east block ofsite. Only representative parts ofthe town wallare indicated.

Entrances are near the centre of the long south­east wall and are usually framed by orthostats. Smallsquare or rectangular box-like annexes severalmetres across are frequently attached to the longnorth-west wall.

The houses vary considerably in size, and in thesample of complete sttuctures, the smallest covers20 and the largest 112 sq. m. The rectangular housesare larger than the 'V-shaped houses (averaging 64and 52 sq. m respectively); houses in the main sitetend to be larger than those of the north-west site (62± 28 versus 52 ± 26 sq. m) but this difference is notstatistically significant. Several of the largestbuildings (all rectangular) seem to include a walledopen space at one end, a feature that also occurs inan excavated house at the late 2nd millennium siteofKharraib (Edens 1999).

Additional small structures are associated with thehouses. These include square structures, roughly 2m across, or more frequently circular sttuclures 1-3

254

m across, which sometimes abut the house wall.These sttuctures resemble the stone-lined hearths orpossibly silos of the kind excavated at Bronze Agesites in Khawlan and elsewhere in the Dbamar area(de Maigret 1990; Wilkinson & Edens 1999).However, not all of these circles belong with theBronze Age town. Some are within houses, fillingthe available space of rooms, and others areembedded in apparently remodelled walls of BronzeAge sttuclures or rest partly on aeolian silts thatpost-date the Bronze Age occupation.

Frequently rough curvilinear walls frameextensive patches of aeolian silts in a terrace-likefashion. Soundings show that the silt of one such'terrace' covers Bronze Age walling and wallcollapse and that similar silts bury occupationdeposits within houses. Such walling thereforeprobably represents a post-Bronze Age remodellingof the site in an effort to retain the silts for agrarianuses. Several curvilinear walls, more carefully

I;IAMMAT AL-QA<: AN EARLY TOWN IN SOUTHERN ARABIA

constructed, extend downslopefrom groups of Bronze Agehouses at the open westernend of the main site; thesewalls may have been fencedenclosures for animals and foroutdoor domestic activitiesattached to the Bronze Ageresidences.

The agricultural systemSince the first field season in1994 it bad been suspectedthat the Bronze Age sites ofthe highlands were associatedwith relict terraced fields.Although it is not possible todate such systems directly, the ijammat al-QAc relictfields fonn a coherent set of features that appear tohave been in use during the Bronze Age occupationof the site. Such an early date for terracedagriculture in the area is not surprising given therecent discovery of a valley bottom terraced fieldburied below some 6 m of valley fill and dated byradiocarbon-dated charcoal from the associated soilto 4960 ± 80 BP (3955-3630 calBC [Beta 117431];Wilkinson 1999: 186).

The agricultural features at ijammat al-Qllc lie onhill slopes that today consist of a veneer of stonydebris, but which originally appear to have beencovered by a thicker humic loam palaeosol. Relictterrace walls appear as horizontal lines of heavilyweathered stones, which are now eroded virtuallyflush with the slope. Excavation in 1996 showedthese to have the characteristic features of fieldterraces, namely a collapsed terrace wall and asupporting mass of fist-size stones immediatelyupslope (Wilkinson 1999: fig. 4). Such relictterraces remain on the south-west, east and north­facing slopes of the hill (Fig. 3). The two last namedgroups are situated near the base of the slope wherethey would have benefited from runoff receivedfrom the large area of hill slope above. In contrast,those terraces on the south-west-facing slope werepositioned half way up the slope, perhaps in partbecause they would have received additional runofffrom the hilltop, thus rendering unnecessary a longcatchment on the slope itself.

FIGURE 6. 3-D reconstruction of Ifammat al-Q4'viewed from the east. First published in Antiquity74.2000.

Additional off-site features include eroded massesof large, heavily weathered stones, 20-25 cm across,which appear to have fonned flattened platfonns onthe north-west and south~t spurs (Fig. 3: Th). Byanalogy with modern installations, and relictfeatures associated with sites of the first millenniumBC and AD, these platforms are interpreted asthreshing floors. They are situated on eminences thatwould receive winds suitable for winnowing grain,and would have been conveniently located for thefields themselves (Fig. 6).

Estimates of site populatioD, poteatial lustaiDlDgareal aad laad use modelThe well defined architecture of l:Iammat al-Qicenables us to make rough estimates of sitepopulation based on the number of visible houses,which in 1999 was estimated to fall in the range 46to 57 (Fig. 7). Each domestic unit appeared toconsist of a nuclear house together with ancillarystructures which included sub-rounded storage unitsand various types of unroofed courtyard space, asdiscussed above. Estimates are made for the mainwalled settlement area, the north-west town outsidethe cross wall, and total site area, that is the sum ofthe two (Fig. 7).

255

TJ. WILKINSON, C. EDENS cl G. BARRATT

Min. no. Max.no. Min. Max. House House

of of pop. pop.S uca uca

houses houses S/house Ibouc Est. I Est. 2

Main town 46 S7 230 28S 2960 3S34

NWtown 16 22 80 110 9S7 1144

Total site 62 79 310 39S 3917 4678

nGURE 7.ljammat al-(!tZC. Population Estimates based on utant architecture.

Pop. 10 IQ

m/person

296

9S.7

391.7

Pop. Est.

Using Est. 2

3S3.4

114.4

467.8

Minimum estimates count the number of distinctand unambiguous housing units, whereas themaximum allows for more latitude in estimates,assuming, for example that certain missing walls canbe reconstructed. If each house is assumed to havebeen occupied by a nuclear or a small extendedfamily of five (Steffen 1979: W127) we getestimates of between 230 and 285 persons for themain town, and 310-395 for the total site area (Fig.7). Alternatively using measured house size andhouse area, and assuming a living space of 10m2 perperson (Kramer 1982), population estimates fall inthe range 2~353 persons for the main town and392 to 467 for total site area. The aboveapproximations are thought to under-estimate thetotal number of houses, because there is evidencefor considerable stone robbing, especially for thenorth-west area. The estimates suggest populationdensities in the range 77-118 personslba for themain town and 62-94 persons! ha for the north-westarea, which are on the low side of populationestimates from elsewhere in the Near East (Sumner1989).

Calculations of potential sustaining area are morehazardous than population estimates because theythemselves are based on estimates of population.Furthermore, estimates have to be made of thecontribution of cereals to the diet and cerealproduction per cultivated hectare, assuming that, uin the Khawlln area, the Bronze Age communitiescultivated a mix of wheat, barley, and perhapssorghum (Costantini 1990). The following

256

provisional estimates are therefore made simply toestablish whether there was sufficient available landfor the settlement, and how that land use may havebeen configured. Today in this heavily populatedlandscape mean landholding size per household isaround 1.4 ba. On the other band, mean rainfedwheat and barley yield is today around 800 kg perba. If mean family size in residence is five personsand cereal requirement is around 250 kg per annum,then, to subsist, each family would require around1.6 ba. For the estimation of Bronze Age sustainingareas we have realistically chosen • slightly moregenerous land holding of 2 ha per household toallow for additional flexibility in cropping patterns.This figure, when applied to the number of housesand estimated population, results in • range of 122­156 ha of cultivated land for the settlement Giventhat the area of land within an arbitrary I km is 314ba, and the total area of the hill around the base is 31ba, • circular area of I km radius would providesome 283 ha of cultivable land (Fig. 8). Becausethere appears to have been some cultivation on thehill slopes and not all land within I Ian was goodcultivable land, we can assume that cultivationprobably spread slightly further into the basin (theqtY) to the south of the site. Clearly however, thearea available on the plain was more than sufficientto provide the cultivation needs oftbe community. Itis therefore significant that the inhabitants cbole tosituate some fields on the hill slopes below the town.preswnably because that formed • convenientlocation for fields.

I,IAMMAT AL-QAc: AN EARLY TOWN IN SOUI'HERN ARABIA

• Village

T Terrace•Site

o

0128

°'05 f1KM.

82l Land > 2400m

~Wadiftoor~"'"

nGUR.E a. Preliminary recOllStnlcJion of the Bronze Age land use around Ifammat a1-~ showing liteinferred location of some of the terraced fields. cultivation on lower slopes and adjacent plain. and valleyfloor pasture.

PrelimiJwy geoarcbaeologica1 studies of valleyfloor relict lake deposits suggests that during themid Holocene (c. 8~3000 Bq the central partsof many basins in highland Yemen held freshwaterlakes or marshes (Wilkinson 1997). Furthermore,there is little evidence for either Neolithic or BronzeAge settlement within such basin centres. Fromthese two lines ofevidence it can be inferred that thebasin centres would have been open space thatremained as grassland or marsh into the third andsecond millennia BC. and thus would have providedvaluable pastoral reserves (Fig. 8). Bronze Age landuse around l;Iammat al-QI(. and in other basins can

therefore be suggested to comprise: a) hill slopesadjacent to the settlement where some cropping andcrop processing took place; b) lower slopes andplain margins devoted to cultivated crops; c) bottomlands of basins which remained as pastoral reserves.

DlscuulouThe perimeter wall, structured fabric andarchitectural density of ijammat al-Ql< justifyca1ling this settlement a Bronze Age town. despiteits relatively small population. Although thesecharacteristics are clearest at ijlDlD1&t aI-QI(. town­sized sites are relatively common in the central

257

TJ. WlLKINSON, C. EDENS &: G. BARRATT

Yemeni highlands. When compared with presentday settlement l:Iammat al-QAc can be seen to fallwithin one of the commonest size classes, namelythe small rural town or large village withpopulations in the range 100-500, of which therewere 11,000 in 1979, holding some 44% of thepopulation (Steffen 1979: 1I/27). Unlike the firstmillennium BC towns of the desert fringe such asShabwa, or Late Bronze Age Sabir on the IndianOcean plain, the buildings at ijammat al-Qlc showvery little variation in size or apparent status. On theother hand the settlements of Khawlln, just 5~Icm to the north-east. make a strong contrast in adifferent way (de Maigret 1990). These sites aresmaller (none exceeding I ha) and containcompounds of contiguous curvilinear rooms setaround a central space; this architectural form seemsto reflect lineage residential groups. The architectureand the significantly drier environment of theeastern highlands suggest that the Khawllnsettlements belonged to communities that placed agreater emphasis on animal herding, perhapsseasonally more mobile and opportunisticallyfarming. However, the evident oorJrasts withijammat al-Qlc may mask similarities of socialorganization, with the linkages between houses andtheir arrangement around a common space withinblocks serving as the seats of extended kin groupswithin the town. These groups probably would havepooled labor in the creation, use and maintenance ofagricultural terraces, threshing floors and otherfacilities.

The settlement system evident in the interior ofsouth-east Arabia provides another expression of asimilar social organization within a differentenvironmental regime. There, settlement wasdispersed, with residential architecture and adjacentgroups of collective tombs clustered around towersthat were scattered through oases. These clusters oftowers, houses and tombs probably containedextended kin groups that coordinated access to landand water, the latter deriving from a combination ofrun-off and ground water. This kind of dispersedsettlement was in fact a recurrent urban feature oflater periods across Arabia (Whitcomb 1996), andwas a common response of kin organization toconstraints ofan arid environment

258

In contrast with the towns of the Gulf coast ofArabia such as QaJCat al-Bahrain, Saar, or TellAbraq, where interaction with other parts of theGulf, Iran. Mesopotamia and the Indus are manifestfrom the artifactual assemblage, those of theDhamIr highlands show little evidence of foreigncontacts, imported materials or trade goods of anysort This suggests that even if the Yemensettlements did not exist in isolation, externalstimulus is unlikely to have been a determiningfactor in their formation. Rather, they appear to havedeveloped in an area which provided excellentpotential for food production, and populationgrowth. These factors do not. however, providesufficient reason for the growth of such towns, andalthough the labour requirement of terracedagriculture, or defense may also have contributed tothe nucleation of population into such settlements,the precise reasons for their growth remains to bedetermined.

NotesI A shorter version of this paper has been

published in Antiqllity (Edens, Wilkinson &Barratt 2(00).

R.efereDcesBlakely J., Vitaliano C. & Brinkman R. 1996.

Bronze Age remains at Kawlah al-Lajarnah.Yemen Arab Republic. Pages 309-333 in W.Overstreet & J. Blakely (cds), EnvironmentalResearch in Support of ArchaeologicalInvestigations in the Yemen Arab Republic,1982-1987. Washington DC: AmericanFoundation for the Study of Man.

Breton J.-F. & Darles C. 1994. Halar Surbin I et 2:villages du Cabal an-NiSIyIn. Pages 46-<i1 in N.Nebes (ed), Arabia Felix: BeitrlJge ZIU' Sprache&UId Ku/tur da vorisJamischen Arabien.Festschrift Walter W. MiUler ZIlm 60.Gelnutstag. Wiesbaden: Harrusowitz.

Cleuziou S. 1982. Hili and the beginnings of oasislife in Eastern ArabiL ~ng! of theSeminarfor Arabian SlIldies 12: 15-22.

Cleuziou S., Inizan M.-L. & Marcolongo B. 1992.I.e peuplement pre- et protohistorique dusystCme fluviatile fossile du Jawf-Hadramawtau Yemen. Paleorienll812: 5-29.

I;IAMMAT AL-QA<: AN EARLY TOWN IN SOlITHERN ARABIA

Costantini L. 1990. Ecology and farming of theprotohistoric communities in the centralYemeni highlands. Pages 187-204 in A. deMaigret (cd.), The Bronze Age Culture ofHawlan al Tiyal and al-Hada. Rome: IsMEO.

Crawford H., Killick R &. Moon J. 1997. TheDilmun temple at Saar, Bahrain and itsArchaeological Inheritance. London: KeganPaul.

de Maigret A. 1984. A Bronze Age for southernArabia. East and West 34: 75-106.

1990. The Bronze Age Culture ofHawlan al-Tiyaland al-Hada. Rome: IsMEO.

Edens C. 1999. The Bronze Age of highlandYemen: chronological and spatial variability ofpottery and settlement. PaIeorient 2SI2: 105­128.

Edens C., Wilkinson T.J. &. Barratt G. 2000.Hammat al-Qa and the roots of urbanism insouthwest Asia. Antiquity 74 (286): 8S4--862.

Frifelt K. 1976. The evidence of a third millenniumBC town in Oman. Jou17lQ1 of Oman Studies 2:S7-74.

H"jlund F. &. Andersen H. H. 1994. Qala 'aJ al­Bahrain. i. (Jutland Archaeological SocietyPublications. 30/1). Aarhus.

1997. Qala 'at ai-Bahrain ii. (JutlandArchaeological Society Publications 3012),Aarhus.

Kramer C. 1982. Village Ethnoarchaeology. NewYork: Academic Press.

Potts D.T. 1990. The Arabian Gulf in Antiquity. (2volumes). Oxford: Clarendon.

1991. Further excavatio1LS at Tell Abraq: theJ990 season. Copenhagen: Munksgaard.

StetTen H. 1979. Population Geography of theYemen Arab Republic. (Beihefte zum TQbingerAtlas des Vorderen Orients). Wiesbaden:Reichert.

Sumner W. 1989. Population and settlement area: anexample from Iran. American Anthropologist91: 631~1.

Tosi M. 1986. Archaeological activities in theYemen Arab Republic, 1986: Neolithic andprotohistoric cultures, survey and excavationson the coastal plain (Tibama). East and West36: 400-41S.

Vogt B. &. Sedov A. 1998. The Sabir culture andcoastal Yemen during the second millenniumB.C. - the present state of discussion.Proceedings ofthe Seminar for Arabian Studies28: 261-270.

Whitcomb D. 1996. Urbanism in Arabia. Arabianarchaeology and epigraphy 7: 38-S I.

Wilkinson T. J. 1997. Holocene environments of thehigh plateau, Yemen, recent geoarchaeologicalinvestigations. Geoarchaeology 12: 833-864.

1999. Settlement, soil erosion and terTacedagriculture in highland Yemen: a preliminarystatement Proceedings of the Seminar forArabian Studies 29: 183-191.

Wilkinson T.J. &. Edens C. 1999. Survey andexcavation in the central highlands of Yemen:results of the Dhamar Survey Project, 1996 and1998. Arabian archaeology and epigraphy 10:1-33.

Wilkinson TJ., Edens C. &. Gibson M. 1997. Thearchaeology of the Yemen high plains: apreliminary chronology. Arabian archaeologyand epigraphy 8: 99-142.

Zarins J. &. AI-Bade H. 1986. Archaeologicalinvestigations in the Tibama plain n, 1985.Atlal 10: 65-107.

Zarins J. &. Zahrani A. 1985. Recent archaeologicalinvestigations in the southern Tihama plain,1984. Atlal9: 65-107.

Contribwors' addressesTJ. Wilkinson, Oriental Institute, 1155 East 58th Street, Chicago, n.. 60637, USA.

email: t-wilkinson@UChicago.cduC. Edens, American Institute of Yemeni Studies, P.O. Box 2658, Sanaa, Republic of Yemen.

email: aiysyem@Y.DelyeG. Barratt, Department of Archaeology, University ofBinningham. P.O. Box 363, Binningham BIS 2IT,

UK. email: Glynn@barra1.demon.co.uk

259