sign in sign up
<<
Home , Agriculture in Israel, Arabian Desert, Ashalim, Desert, Desert climate, Desert farming

ENVIRONMENT AND IN THE CENTRAL AND -BARNEA DURING HISTORICAL TIMES PROMOTOREN: dr. ir. L. J. Pons hoogleraari nd eregional e bodemkunde

dr. A.M. vande rWoud e hoogleraar in deagrarisch e geschiedenis Hendrik J. Bruins

DESERT ENVIRONMENT AND AGRICULTURE IN THECENTRA L NEGEV AND KADESH-BARNEA DURING HISTORICAL TIMES

Proefschrift ter verkrijging van de graad van doctor in de landbouwwetenschappen, op gezagva n de rector magnificus, dr. C.C. Oosterlee in het openbaar te verdedigen op vrijdag 23me i 1986 des namiddags te vier uur in de aula van de Landbouwhogeschool te Wageningen.

1986 I-mm- MIDBARFoundatio n /Nijkerk , TheNetherland s

\ i • . °\ Copyright (c) 1986b yHendri kJ .Bruin s

All rights reserved. Nopar t ofthi spublicatio nma yb ereproduced , stored ina retrieva l system, ortransmitte d inan yfor mo rb yan ymeans , electronic, mechanical, photocopying, recording, or otherwise, without theprio r written permission fromth eMidba r Foundation.

ISBN 90-71666-01-8

CIP-GEGEVENSKONINKLIJK E BIBLIOTHEEK,DE NHAA G

Bruins, HendrikJ .

Desert environment andagricultur e inth ecentra l Negev and Kadesh-Barnea during historical times/ Hendri kJ . Bruins. -Nijker k :Midba r Foundation.- Fig. , foto's,tab . Proefschrift Wageningen. -Me tlit .opg .- Me tsamenvattin g inhe tNederlands . ISBN 90-71666-01-8 SISO63 5 UDC 63(252)(091)(043.3) Trefw.: landbouw ;Nege v ;geschiedeni s/ landbou w; Kadesh-Barnea ;geschiedenis .

Copieso fthi s dissertation mayb eobtaine d from either

MIDBAR FOUNDATION or H.J.Bruin s P.O.Bo x 78 Jacob Blaustein Institute forDeser t Research 3860A BNijker k Ben-Gurion University ofth eNege v The Netherlands SedeBoqe r Campus, 84990

Price: DF1 39.95 (Dutch guilders);DF 1 4.95 - postagean dpacking .

Payment eithero nth eban k accounto fth eMidba rFoundation : 37.78.34.009 - Rabobank,P.O .Bo x44 ,386 0A ANijkerk ,Th eNetherlands ; or enclose remittancewit h order; postal ordersan dcheque s shouldb e made payablet oth eMIDBA R FOUNDATION.

Cover: The ofEi ne lQudeira t (Kadesh-Barnea)i n the north­ eastern Sinai desert; noteth enorther n wallo nGebe le lQudeira tan d the central Negev mountainsa tth ehorizon .

BIBLIOTHEEK DER iiSDBOtJWHOGESCHOO* VAGENINGEN AiA/0£?QI, ,0\> J

1 •' t• ~ K STELLINGEN i-i-U;i.oJ- S ttOGESCHOOl WAGENINSEN

1. About one third of the 's terrestrial surface is occupied by dry , usually divided into hyper-, arid and semi-arid zones. Runoff farming isth e only form of agriculture based upon local rainfall which is possible in the arid zone senso stricto, besides extensive livestock rearing.

2. Runoff farming or rainwater-harvesting agriculture is farming in dry by means of runoff rainwater from whatever type of catchment or ephemeral . H.J. Bruins, M. Evenari, U. Nessler (1986) Rainwater-harvesting agricul­ ture for food production in arid zones: the challenge of the African . Applied Geography 6:13-32.

3. The following runoff farming systems are distinguished, arranged in order of increasing hydro-geomorphic scale: 1)Micro-catchmen t system 2) Terraced system 3)Hillsid e conduit system 4) n system 5)Diversio n system A thorough understanding of the is required, in terms of geology, , soils, and hydrology, to evaluate the suitability of an area for runoff farming and to design the proper system in each particular situation, taking socio-economic factors into account. H.J. Bruins, M. Evenari, U. Nessler (1986) Rainwater-harvesting agricul­ ture for food production in arid zones: the challenge of the African famine. Applied Geography 6:13-32.

4. Pastoral nomadism isdoome d to stagnation, because increased production in nomadic society asa whol e is,t o any significant extent, impossible. Khazanov, A.M. (1985:71,76) and the outside world. Cambridge University Press.

5. Combinations of semi-extensive livestock rearing and runoff farming ought to be considered in the arid zone of developing countries to increase local food production and to enhance system viability.

6. The promotion of agrarian self-sufficiency in developing countries suffering from malnutrition and famine should be the main policy guideline in agricultural development cooperation.

7. Internal food reserves must form an integral part of self-sufficient agricultural systems in arid regions,a s abuffe r for the inevitable years ofdrought . 8. Instrumental recordso fweathe r observations indicate that from about195 0 extreme variations inclimati c patterns began tooccu r inman y parts of the world. Agricultural planners ought totak e this into account. Buffer stockpiles offoo d should beforme d invulnerabl e agro-climatic regions. Diversification of agro- should bestimulate d to lessen the impact ofclimati c shocks,a sth eone-cro p economywa sa tth eroo to fman y ofth egreates t inth epast . Partly based on Lamb, H.H. (1982) Climate, history and the modern world. London: Methuen.

9. There appears tob ea relationship between themajo r historic period of sediment accumulation inth eKadesh-Barne a valley, from A.D. 1200-1700, and climatic conditions associated with theso-calle d Little .

10. Experimental archaeology can make an important contribution to the understanding ofth epast .

11. Geologists and archaeologists should beware of the "reinforcement syndrome" by whichne wdat a areforce d into an existing stratigraphic framework without evaluating thevalidit y ofth eframewor k inth eligh to f thene wdata . Modified after Watkins, N.D. (1972) Reviews of the development of the geomagnetic polarity time scale and discussion of prospect for its finer definition. Geological Society of America Bulletin 83:551-574.

12. "Thestead y onward flowo ftime , which isth eessenc e ofth ecause-effec t relation, issomethin g whichw esuperimpos e ont oth eascertaine d lawso f nature outo fou row nexperience ; whether orno ti ti sinheren t in the nature oftime , wesimpl y dono t know". Jeans, Sir James (1930:32) The Mysterious Universe. New York: Macmillan.

13. Tounderstan d aphenomeno n wehav e notonl y tokno w what itis , but also how itcam e into being. Boas, F. (1940:305) Race, Language and Culture. New York: Macmillan.

14. Without the unigue food provisioning of , theancien t would have perished during their sojournment inth ehyper-ari d to arid Sinai desert,afte r theExodu s from .

H.J. Bruins. Desert Environment andAgricultur e inth ecentra l Negev andKadesh-Barne a during historical times. Auditorium, Agricultural University ofWageningen ;2 3Ma y 1986. ABSTRACT

Bruins, H.J. (1986) Desert Environment and Agriculture in the central Negev and Kadesh-Barnea during Historical Times. Doctoral dissertation,Agricultura l University ofWageningen ,Th e Netherlands. XI + 219 p.,2 7 tables,2 9 figures,4 8photos ,22 8 refs,2 app.Dutc h summary. use based on local rainfall in the arid zone senso stricto is often limited to pastoralism, sometimes combined with very marginal rainfed farming, unsuccessful inmos t years. A more sophisticated form of rainfed agriculture - runoff farming -ha sbee n practised in the central Negev and adjacent north­ eastern Sinai by a sedentary population incertai n historical periods, parti­ cularly during Byzantine times from the 5th to 7th century A.D. The environ­ ment ofth eRunof fFarmin g District in theNege v isdescribed , aswel l as the mechanics of rainwater-harvesting agriculture or runoff farming. Five systems of runoff farming aredistinguishe d on hydro-geomorphic criteria. Excavations have been carried out in ancient runoff farming wadi terraces at Horvat Haluqim, apparently dating back some three millennia, and inNaha l Mitnan, attributed to theLat e Byzantine -Earl y Arab period.Th e practising of runoff farming at Horvat Haluqim in antiquity has been substantiated by specific soil development in an ancient wadi-terrace layer, indicative of periodically flooded conditions. Calculations have beenmad e about food production, based upon yields, in relation toestimate d population levels in the past. Besides runoff farming, oasis- agriculture in the could only be practised at aver y few spots. The valley oasiso fKadesh-Barne a or Ein el Qudeirat, situated in northeastern Sinai, ison e ofth e most outstanding places in this respect inth e entire Sinai-Negev desert, endowed with a copious and cultivable soils.Th e systems of irrigation agriculture are described, whilst theremnant s of ancient aquaducts have been dated by radio­ carbon.Th e valley stratigraphy and soil development have been investigated in detail, in relation toth e tell (mainly composed ofancien t Israelite fortres­ ses, although theEarl y Fortress might date back toth eLat e , as suggested by C-14dates) , and in relation toancien t remnants of irrigation agriculture, dated by radiocarbon to theBronz e age and the 7th century A.D. Dramatic cut-and-fill processes have occurred inth e Kadesh-Barnea valley during historical times. These remarkable changes have been substantiated, in addition to the stratigraphic evidence,throug h the detection of specific soil development, particularly themicroscopi c discovery ofauthigeni c pyrite. A detailed chrono-stratigraphy has been established based upon radiocarbon dates, determined by the Isotope PhysicsLaborator y ofProf .Moo k (University of Groningen). All the radiocarbon dates have been calibrated from conventio­ nal C-14 years intohistorica l years. The carrying capacities of the Runoff Farming District and theKadesh-Barne a -Quseim a region have been calculated and the outcomes arecompare d with former population levels.Th e relationships between the landscape, climatic and agricultural history are evaluated. A clear time-correlation exists between valley aggradation and a relatively wet climate, as asbetwee n valley incision and a relatively dry climate. There is a bit of irony inth e conclusion that runoff farming was not prac­ tised by a sedentary population in the central Negev desert in those periods during the last threemillenni a when the climate was relatively wet. Internal and/or external food reserves must have been an integral part of a sedentary runoff farming society, asa buffer for the inevitable yearso f . Key words: historic land-use ina n arid desert,deser t environment,ari d zone farming, ancient agriculture, runoff farming, environmental archaeology, paleo-climate/landscape/land-use relationships, soil development, geology. Aan Willie,

Esther,Naomi ,e nIri t

"OTD3 lap Tbiob AT :•- -\ IJ- :

Tohi mwhic hle dhi speopl e throughth edeser t

(Psalm 136:16)

VI ACKNOWLEDGEMENTS

First Isincerel y thank my promotors,professo r Dr.L.J . Pons and professor Dr. A.M. van der Woude, for their interest in the subject, their continuous support and forman y helpful suggestions inth e various stages toward the completion of this research and the writing of the dissertation. I am indebted toMr .Rudolp h Cohen, District Archaeologist of theNege v of the Department ofAntiquitie s and Museums, aswel l as to the Archaeological Survey of Israel, for the opportunities Ireceive d tocarr y out field-work in the Negev desert and the area ofKadesh-Barnea . During these periods I got acquainted with many interesting aspects of ancient man-land relationships in this arid region. The various discussions with Mr.Rudolp h Cohen, aswel l as with Mr.Mordecha i Haiman, Mr.Yose f Porat, Mr.Do v Nachlieli, and others contributed to shape my archaeologic-historic perception ofth e area. Living and working in the Negev desert at the Jacob Blaustein Institute for Desert Research (Ben-Gurion University of theNegev )ha sconsiderabl y enriched my understanding of this unique environment in various aspects. The coopera­ tionwit h professor Dr.Michae l Evenari, Mr.Davi d Mazigh andMr .Arie h Rogel enabled me to obtain abette r comprehension of runoff farming. The investiga­ tions with professor Dr.Arie h Issaran d Mr.Arno nKarniel io n Pleistocene landscape-climate relationships in the Sede Zin area provided a useful back­ ground inth e interpretation ofLat eHolocen e landscape developments at Horvat Haluqim. I thank many ofm y dear friends and colleagues ofth e Institute for their encouragement and support.Th ecomment s by professor Dr.Loui sBerkofsk y and professor Dr. on certain parts ofth e text are very much appreciated. Without theman y radiocarbon dates determined at the Isotope PhysicsLabo ­ ratory ofprofesso r Dr.W.G .Mook ,Universit y ofGroningen ,th e absolute basis of stratigraphy - time -woul d have been lacking inth e investigation of the fascinating environmental history of theKadesh-Barne a valley. TheC-1 4date s proved to be very important in the framework ofthi s dissertation. Hence Ia m indebted to professor Mook for hiswillingnes s to cooperate in this research, and for his comments onpart s of thetext . I thank the following persons connected with The Hebrew University of Jerusalem: Dr.Pau lGoldber g (Institute of Archaeology) for introducing me in the Kadesh-Barnea area in 1976; Dr.Ra n Gerson (Institute ofEart h Sciences) and Dr.Yora m Tsafrir (Institute of Archaeology) for useful discussions in the

VII beginning stages ofthi s research; Professor Dr. Dan Yaalon (Institute of Sciences) asm y former supervisor and mentor who introduced me into the desert of the Negev and taught me to "read" the landscape in relation to its environment. My thanks to Dr.Chai m Benjamini (Ben-Gurion University of the Negev, Department ofGeolog y and Mineralogy) for hiscomment s oncertai n geological parts ofth etext . I very much appreciate the support received from theDepartmen t of Soil Science and Geology,Agricultura l University ofWageningen ,wit h regard to the laboratory aspects of this research. The contribution ofth e following persons of this Department is acknowledged: - Ir.R . Miedemaan d Mr.A.G . Jongmans for their helpful suggestions in the microscopy study of the thin sections. - Mr.O.D .Jeronimu s for the preparation ofth e thin sections. - Mr.L.Th . Begheijn and Mr.F.J . Lettink for thechemica l analyses. - Prof.Dr . S.B.Kroonenberg , Dr.N . vanBreemen ,an dDr .R . Brinkman for useful comments. - Mr. Z. van Druuten for the photographic processing of some colour slides and microscope pictures.

The comments of Ir.A .Kamphors t on aspects of soil salinity and of Ir. P.D.J,va n der vorm on soil fertility (Agricultural University of Wageningen, Department ofSoil s andFertilizers ) are appreciated. The discussions with Dr.Hele n Scoging and Mr.Michae lLe e of the London School of Economics about hydrology and ancient runoff farming have been useful and stimulating. I acknowledge theassistanc e ofMr .B .Goudswaard , Mr.Aharo n vanLeeven , Mr. Benny Peleg, Mr.J . Meier, Mr.J . Heijenga,an d others,i n certain parts of the fieldwork. My thanks to theBiali k Foundation in Jerusalem for their permission to use and re-edit certain parts from chapter 8 of the book (inHebrew) : The Ancient Agriculture in theNege v (1967), by Y.Kedar .Thi s material appears primarily inchapte r 2,p . 11-17. Finally, Ia m very grateful for the continuous encouragement and support of my wife Willie and my three daughters, Esther,Naom i and Irit,wh o all helped in various ways, relieving me ofsom e family duties, which enabled me to concentrate fully on thewritin g of this dissertation.

VIII CONTENTS

Abstract V Acknowledgements VII 1 INTRODUCTION 1 Food production anddrought s 1 Classification ofdr yland san dland-us e 3 The central Negev desert andth evalle y oasiso fKadesh-Barne a 5

2 ENVIRONMENT OFTH ERUNOF F FARMING DISTRICT INTH ENEGE V DESERT 9 General Setting 9 Geography ofth erunof f farming district 11 Geology andGeomorpholog y 17 Climate 22 Soilsan dVegetatio n 25 Brown lithosolso nhillslope s ofhar d calcareous rocks 27 Calcareous desert lithosolso nhillslope s ofsof t calcareous rocks 29 Young alluvial soilsi nth eactiv e stream channels 29 Loessial serozems 30 Sandy soils 31

3 RUNOFF FARMING INTH EREGIO N 33 Introduction 33 Terminology ofrainwater-harvestin g agriculture orrunof f farming 34 History ofrunof f farming inth eNege v 36 Chalcolithic 36 Bronzeag e 37 Israelite period 37 Late Nabatean-Roman-Byzantine-Early Arab period 37 farming 38 Modern Israeli runoff farming 38 Hydrology ofrunof f farming 38 Runoff farming systems 45 Micro-catchment system 46 Terraced wadi system 47 Hillside conduit system 49 Liman system 50 Diversion system 51

4 PASTORALISM 55 Nomadic pastoralisma sa specialize d formo fa food-producin g economy 55 Pastoralisman dagricultur e 57 The originso fpastoralis m 59

5 HORVAT HALUQIM 62 Environment 62 Geomorphology andgeolog y 62 Climate 63 Hydrology 65 Soilsan dvegetatio n 65 Archaeology ofHorva t Haluqim 67 Water supply inth epas t 70 Ancient runoff-farming wadi terracesa tHorva t Haluqim 71 Stratigraphy andsoi l development ofa runoff-farmin g wadi terrace 75 The discovery ofa nancien t terrace layer 75

IX Laboratory data 79 Soilmicromorpholog y 82 Stratigraphic conclusions 84 6 ESTIMATED WHEAT YIELDSUNDE R RUNOFF-FARMING CONDITIONS INANTIQUITY , 86 RELATED TO HORVATHALUQI M Introduction 86 Wheat yieldsbase d on ancient literary sources 87 The quantity ofwhea t sown in relation to soil and climate 89 Estimated amount ofwhea t sown inth e Negev and corresponding yields 90 Evaluation ofancien t wheat yields inth e light ofmoder n research 91 Wheat yields from theAvda t experimental runoff farm 92 Estimated wheat yields and population levels atHorva tHaluqi m 94 7 NAHALMITNA N 97 Environment 97 Archaeology 98 Stratigraphy and soil development inByzantin e wadi terraces 99 Stratigraphy 99 Soil development 101 8 KADESH-BARNEA: ENVIRONMENT,HISTOR Y AND THE TELL 105 Environment 105 Geomorphology and geology 105 Geo-hydrology 108 Agricultural evaluation ofth e spring water 109 Rainfall 109 Soils and 110 History and geography 110 Archaeology ofth etel l 112 Carbon-14dat e (GrN-12330)- Lates t Bronze toEarl y Age 112 Early,Middle ,an d UpperFortres s 116 Persian period 117 Environment and siting ofth e tell 117 9 KADESH-BARNEA: IRRIGATION AGRICULTURE 121 Systems of irrigation agriculture in theKadesh-Barne a valley 121 Wild flooding system 123 Border irrigation system 124 Basin irrigation system 125 Archaeologic and historic evidence of irrigation agriculture 125 Bronze age aquaduct:Carbon-1 4 date (GrN-12327) 125 Iron age 128 Byzantine period 128 Carbon-14 dating ofth e dam (GrN-12326) 130 Mortar micromorphology and paleo-environmento fth e dam's aquaduct bottom 132 Authigenic pyrite 132 Carbon-14 dating ofa n aquaduct beginning inth ewad i bed 134 Mortar micromorphology ofth e 7th century wadi-bed aquaduct 137 Authigenic pyrite 139 Bedouin irrigation farming inth eKadesh-Barne a valley 139 10 KADESH-BARNEA: VALLEY STRATIGRAPHY AND SOIL DEVELOPMENT 144 The valley section near theLat e Byzantine -Earl y Arab dam 144 Thechanne l bluff profile that exhibits twoC-1 4date d aquaduct remnants (Bronze age andLat e Byzantine -Earl y Arabperiod ) 151 Valley aggradation and its radiocarbon dating near the tell 154 Soil development 161 Micromorphology and paleo-environmento fKB-12 9 162 Authigenic pyrite 162 Micromorphology and paleo-environment ofKB-13 1 165 Authigenic pyrite 165 Micromorphology and paleo-environment ofKB-13 2 166 Granulometry and chemical analysis ofth e aggradational deposits south ofth etel l 167 An excavated soilpi twes t of the tell 170 Stream channel evolution in theKadesh-Barne a valley during historical times 172

11 WATER,FOO D PRODUCTION AND POPULATION INTH EKADESH-BARNE A AREA 175 Introduction 175 Estimated maximal food producing capacity and related population levels ofth eKadesh-Barne a valley in antiquity and today 176 Comparison ofestimate d food production and related population levelswit h archaeologic data ofLat e Byzantine-Early Arab times 180 Remainso frunof f farming inth evicinit y ofKadesh-Barne a 181 The south of theKadesh-Barne a valley 181 Gebel elQudeirat ,nort h ofth eKadesh-Barne a valley 182 Gebel elEi n 182 Wadi elHaluf i 183 The western part ofGiva t Barnea 183 The post- sojournment ofth e ancient Israelites in the desert 186

12 RELATIONSHIPS BETWEEN THELANDSCAPE ,CLIMATIC ,AN D AGRICULTURAL HISTORY OF THEREGIO N 188 Valley changesi nrelatio n toclimat ean d agriculture 188 Introduction 188 Theclimati c relationship 189 Roman filldeposi t 189 Incision during theEarl y Arab period 189 Main filldeposi t (LateCrusader-Mamluk-Earl y Ottoman period) 190 Incision since theLat e Ottoman period untiltoda y 190 Assessment ofpossibl e anthropogenic influences 191 Reflections about the relation between climate and valley history 192 Climatic variations and the viability of runoff farming 194 Paleo-climatei nth eregio n during historical times 194 Relations between climate and runoff farming 194 The frequency ofdrough t years 196 and the viability ofrunof f farming 198 Food production and population in theRunof fFarmin g District 200

APPENDICES 1.Archaeologica l and historical periods in Israel,wit h emphasiso n theNegev . 203 2.Laborator y methodso f sediment and soilanalysi s 204

REFERENCES 205

SAMENVATTING 215

CURRICULUM VITAE 219

XI 1 Introduction

FOOD PRODUCTIONAN DDROUGHT S

About one thirdo fth eworld' s terrestrial surfacei s occupied by dry lands, situatedi nnearl y7 0countries .Peopl e livingi ndr yregion s face special problems toproduc e enough foodi na high-ris k environment from an agricultural pointo fview . Thisi sexemplifie d byth econdition so fdrough t and famine that have struck partso fth eAfrica n duringth e seven­ ties andeighties , causing suffering forman y people. There appears tob e growing evidence that weather fluctuations have increasedi n frequency and amplitude during thelas ttw odecades , resultingi ngreate r instability in localan d world food supplies (Abele tal. , 1981;Lamb , 1982). The study ofth e impacto fclimati c variationsan dchang e upon landscape and agriculturei ndr yregion s deserves attentioni nvie wo fthes e problems. It is necessary toinclud e thepas ti nthi s fieldo f research, apart from intrinsic archaeologicalan d historical objectives.Thos e involvedi nagricul ­ tural planningan dfoo d production needt eawar eo fth e possible frequency and amplitudeo ffluctuation si nth e weatheran dclimat ei na certai n region. The experienceo fth epresen t generationi nterm so fclimati c changean dit s effect upon landscapean dagricultur e only coversa relativel y short span of time. Climatic patterns in the pastma ywel l have been more radical and extreme than those within present memory,and ,b yimplication ,ma yb es o inth efutur ea swell .A reasonabl e assessmento fthi samplitude ,th efrequen ­ cyo fvariation ,an dit s likely impacto nenvironmen tan dagricultur e can only be established when the factor timei sgreatl y extended beyondth e present to includeth e past.

1 Whereas interactions between climate and landscape are to a significant extent not in the realm ofhuma n control and influence, the issue of food production and food provision ismor e than just the passive result of physical environmental factors upon human activity. It is,therefore ,importan t also to study the question of human adaptation or lack of adaptation to climatic variations or change. The value of this field ofendeavou r is ofmajo r rele­ vance to the problems ofmoder n world planning, as described in an excellent article by Ingram, Farmer and Wigley (1981): "It isplai n that inorde r to advance the study ofhuma n adaptations toclimati c variations to the point where the findings may beo f real use tomoder n planners,i twil l be necessary to try to specify more clearly what degrees and types of climatic stress impose the greatest problems of successful response, and to seek to identify more rigorously the key features ofmor e or less adaptive societies". An interesting investigation in this direction was made by Bowden at al. (1981)i n arid and semi-arid regions.The y examined material on agriculture in the US Great in theperio d A.D. 1880-1979, on droughts in the region of during 1910-1915 and 1968-1974, and on the societies of the - Valley from 6000 BP to thepresent . The complementary hypo­ theses are discussed that, over time, societies adapt to cope with 'minor' climatic stresses, but that, thereby, they do little to decrease, and may actually increase,thei r vulnerability to 'major' stresses of rare fequency. Both in the Sahel and theAmerica n Great Plains theke y factor in reducing vulnerability appeared to be the integration of each area into a wider econo­ mic system. However,Bowde n et al.war n that there may prove to be a limit to the effectiveness ofsuc h integrative mechanisms in the future. More climatic stress in the Sahel, coupled with other factors, could precipitate system collapse in this region with an already fragile resource base. Mooley and Pant (1981)collecte d data on the 32principa l droughts which affected during the period 1771 - 1977. These droughts resulted in famines, causing hardship to a large number of people.Statisti c tests showed the droughts to occur randomly in time, following aPoisso n probability model of occurrence in a 5-o r 10-year period. These droughts caused food shortages, whereby prices of available food rose beyond thepurchasin g power of the rural population, pushing the farmer deeper in debt and leading to both exploitation and riots. The authors advocate the building up of food grain reserves in rural areas and special funds to afford more protection against drought events. The question of human adaptation to climatic variation is a complex one, whose,systemati c study has only just begun. Although it isunealisti c to draw simple parallels with former situations and societies, it is reasonable to suppose that some useful lessons might be learned from past events.According ­ ly,politician s and planners have begun to show a lively interest in identify­ ing and measuring the effect ofclimati c fluctuations and changes on past societies (Ingram et al., 1981).

CLASSIFICATION OF DRY LANDS AND LAND-USE

Dry lands are generally classified into three major groups or zones: hyper-arid, arid, and semi-arid regions.Althoug h ranked together,th e large differences in environment and land-use between each ofthes e dry zones cannot be emphasized strongly enough. Unesco (1979)ha s developed a useful system to delimitate hyper-arid, arid, and semi-arid regions on the basis of measurable data expressed as aridity indices.Th e degree of bioclimatic dryness or aridi­ ty depends on the relative amounts ofwate r gained from rainfall and lost by evaporation and . Bioclimatic aridity can thus be expressed in terms of the ratio P/ETP, inwhic h P is themea n value ofannua l and ETP the mean annual . Themarke d differences between the dry zones are shown inTabl e 1. This threefold division of the dry lands largely coincides with historic land-use development in these regions. It should beemphasize d that only the semi-arid zone issuite d for normal rainfed agriculture and sedentary live­ stock rearing. The hyper-arid zone corresponds to the true or inner desert,a deserted no-man's land without any agricultural or pastoral land-use, except in oases or along exotic rivers. The term arid zone needs some clarification to avoid semantic entanglement, because this term isals o used in the litera­ ture to denote the three defined dry regions together. Inthi s dissertation the term isgenerall y used in its narrow, defined sense.Th e arid zone, senso stricto, is traditionally a 's land, practicable for nomadic livestock rearing,bu t too dry for common rainfed farming. There exists, however,a specialized form of farming based upon the use of runoff water generated by local rainfall,whic h is suitable for the arid zone: rainwater-harvesting agriculture or runoff farming.Thes e synonymous terms are used interchangeably and defined as farming in dry regions by means of runoff rainwater from whatever type of catchment or ephemeral stream (Bruins etal. , 1986). Runoff collection may also be used in the semi-arid zone to improve the 1. Classification, characteristics and land-use of the dry zones (largely based upon information from Unesco, 1979)

HYPER-ARID zone: -P/ET P ratio is smaller than 0.03 - annual rainfall isver y low - interannual rainfall variability up to 100 % - very sparse vegetation - norainfe d agriculture or grazing

ARID zone: -P/ET P ratio ranges from 0.03 - 0.20 - annual rainfall 80 - 150m m inwinte r rainfall areas 200 - 350m m insumme r rainfall areas - interannual rainfall variability 50-100 % - scattered vegetation - nomadic livestock rearing is possible - agriculture based upon local rainfall only feasible through rainwater-harvesting techniques:runof f farming

SEMI-ARID zone: -P/ET P ratio ranges from 0.20 - 0.50 - annual rainfall 200- 500m m inwinte r rainfall areas 300 -80 0m m insumme r rainfall areas - interannual rainfall variability 25-50 % - discontinuous vegetation with pe^rennial grasses - rainfed agriculture and sedentary livestock rearing are common

viability of rainfed farming (Huibers, 1985). There isa fundamental diffe­ rence, however,betwee n the use of runoff in arid and semi-arid zones. In the former zone the amount ofdirec t rainfall isno t sufficient inmos t years to sustain a crop. By collecting and storing the runoff rainwater of a certain landscape catchment into the soil of a smaller cultivable area, the actual amount of concentrated water in this latter area ( + runoff) may be adequate to practise agriculture. Inth e semi-arid zonedirec t rainfall is sufficient inmos t years to raise a crop.Runof f generated in the agricultural fields themselves,no t necessarily from adjacent non-arable catchments,ma y be stored in shallow ponds asa source ofadditiona l irrigation water in times of moisture stress (Huibers, 1985). Thus the use of runoff in semi-arid zones increases the viability of rainfed farming. Inari d zonesrunof fwate r is absolutely essential topractis e agriculture based upon local rainfall, the amount of runoff conducted unto the agricultural fieldsbein g much larger than the amount of direct rainfall itself. As local food production within the arid zone ought to increase,especiall y in rural or remote areas, rainwater-harvesting agriculture seems tohav e a largely untapped potential in this respect (LeHouero u and Lundholm, 1976). Moreover,ther e isa limi t to the useo fextraterritoria l river water or good- quality inth e arid zone for irrigation purposes. Regions irriga­ ted with water from exotic rivers like the , Tigris,Euphrates ,an d Indus are the exception rather than the rule. Rainfall is often the only available water source forman y arid regions. However, rainfall amounts and its intra-annual and inter-annual variability (Berkofsky, 1984) must be studied carefully toasses s the suitability of a certain arid region for rainwater-harvesting agriculture. Inarea s of similar mean annual precipitation, regions of greater rainfall variability are less viable from an agronomic point ofview . Suitability ofa regio n for runoff farming not only depends upon climate, but also very much upon landscape and soil properties. A sandy or landscape isusuall y adisadvantag e in this respect.

THE CENTRAL NEGEV DESERT AND THE VALLEY OASISO F KADESH-BARNEA

This dissertation deals with the interactions ofclimate , landscape and agriculture, during historical times, inth eari d region of the central Negev hills and the valley oasis of Ein el Qudeirat orKadesh-Barnea , situated in an adjacent part ofnorth-easter n Sinai. Theancien t meaning inHebre w of thewor d "negev" isdryness , which isver y appropriate indeed. TheNege v and Sinai form part of the largest planetary desert belt on earth, extending from the Atlantic ofSahara n west Africa unto India. Within the division of dry zones, aspresente d in Table 1,th e hills of the central Negev and north­ eastern Sinai belong toth e arid zone senso stricto. The present meanP/ET Paridit y index for thisregio n isabou t 0.07. This is quite a low figure, which shows that the centralNege v hills are situated at the dry edge of the arid zone, close to the hyper-arid zone of the inner desert. Onemigh t expect thatmargina l nomadic livestock rearing would be the 35- -35

30- -30

25- •25

I 40

FIGURE 1. The location of the Negev and Sinai as aland-bridg e between Africa and . sole type of land-use possible in such adr y region. The fact of the matter is, that runoff farming based upon local rainfall wascarrie d out here exten­ sively in time and space (Kedar, 1967;Evenar i et al., 1971, 1982). Orthodox irrigation agriculture from apermanen t water source,o n the other hand, could only be practised at aver y fewplace s in this entire region during former times. TheKadesh-Barne avalle y innorth-easter n Sinai consti­ tutes the main centre in this respect, due to the rareexistenc e of a spring with copious,good-qualit y water,an d adjacent cultivable land. Inconclusio n ofthi s introduction abrie f sketch of the following chapters is hereby presented. The desert environmentan dgeograph y of the runoff farming district are described inconsiderabl e detaili nth e second chaptero f this dissertation. Themechanic so frunof f farming,it smai n regional history, as wella shydrologica lan dgeomorphi c aspects, are presented inchapte r 3, which includesa descriptio no fth e different runoff farming systemsi nrela ­ tiont olandscap e properties (Bruinse tal. , 1986a). Pastoralism orextensiv e livestock rearing has, invariou s forms, often played animportan t rolei nth eari d zone. Its different systems have been discussed in chapterA i na somewha t wider regionalan dhistorica l perspec­ tive, because remains ofthi s typeo fland-us e cannot alwaysb e traced in archaeologic and historic terms withth esam e kindo f detail as sedentary runoff farmingo roasi s agriculture. Ancient man-land relationships concerning runoff farming were studied at the site ofHorva tHaluqi m (chapter 5), apparently dating back some three millennia (Cohen, 1976), and inNaha lMitna n (chapter 7)o fapparen t Late Byzantine origin (Haiman, 1982).Th esedimentar y andarchaeologica l stratigra­ phyo fancien t runoff farming wadi terraces were investigated, aswel la sth e related Late Holocene soil development. The potential food production of runoff farming inantiquity , calculatedo nth e basiso f wheat yields, is discussed in chapter 6an drelate dt oth esit eo fHorva t Haluqim andit s estimated former population. A detailed studyo fth e Kadesh-Barnea areai nchapter s8-1 1 revealed quite astonishing valley changeswit h unusual soil development during historical times. Archaeological remainsan dradiocarbo n dates from charcoal provided valuable stratigraphic time markers. There seemsa possibility , asindicate d by someo fthes e carbon-14dates , that irrigation agriculture with water from the local spring was already practised inth eKadesh-Barne a valley duringth e Middle orLat e Bronze age, whilst theEarl y Fortressmigh t perhapsb e older than hitherto assumed (Cohen, 1980,1981 ,1983) . Potential food production levels inth e Kadesh-Barnea valleyar ecalculate dan dcompare d with estimated population levelsi nantiquity , wherebyth e post-Exodus sojournment ofth e ancient Israelitesi nth e deserti sals o touched upon. The relationships betweenth elandscape , climatican dagricultura l history of the central Negev desertan dKadesh-Barne a are discussedi nth efina l chap­ tero fthi s dissertation.Th eviabilit yo frunof f farmingi nsuc ha dr yregio n is evaluated. Possible food production levelso fth e runoff farming district in antiquity are calculated andcompare d with estimated former population numbers. = wadi = ephemeral stream 7 FIGURE 2.Th eNege v and the location ofth e runoff farming district. Environmento fth erunof f farming districti nth e Negev desert

GENERAL SETTING

Although the Negev constitutes onlya tin y parto fth e hugeSaharo - belt, itca nb eregarde da son eo fth emos t famous in the world. Itsfam e is, howeverno ts omuc h based upon unique geographical attributes, butdu et oit s connection with the Booko fbooks .Th efirs t times the Negev ismentione d inth e Scriptures (Genesis1 2an d13 , quoted below) suggest itsenvironmenta l qualitiesa sa regio n suited forpastora l livestock rearingi nth e Middle Bronze Age (around 1900 B.C.), for usedt osta y inth e area with his flocks. However, whena famin e struckth eland , Abraham wentt oEgypt . It seems likely thatth efamin e was causedb y a drought, typical for marginal dry areas. Egypt often provedt ob ea regiona l placeo frefug e in sucha situation . Itsagricultur ean dlivestoc k productionar edependen t upon theNil e water, fedb yrain si ntropica l Africa and,therefore ,unaffecte db y droughts over theLevant . Another important geographic aspecto fth eNege vi s its role, together with Sinai, asa natura l land-bridge between Africa and Asia. "And Abram journeyed, goingo nstil l toward theNegev . Andther e was a faminei nth eland ; andAbra m went down into Egyptt osojour n there; forth e famine was grievousi nth e land...an dh eha dsheep ,an doxen ,an dh easses,. , and she asses, andcamels " (Genesis 12:9,10,16). "And Abram wentu pou to f Egypt, he,an dhi swife , andal l thath ehad , and with him, into the Negev.An dAbra m was very richi ncattle... " (Genesis 13:1,2). The precise limits ofth e Negevar edisputabl e in geographical terms. Especially in a southwesterly direction, there is no clear natural boundary separating the Negev from the Sinai, except perhaps Wadi el Arish and its tributary Wadi Quraiya.Th e political boundary between theNege v and Sinai was drawn in 1906 as theresul t ofa n agreement between Britain and . Thus the Negev and Sinai were demarcated along amor e or lessstraigh t line running from theMediterranea n coast near unto theRe d Sea just south of Elat. Today this line constitutes the border between Israel and Egypt. The eastern boundary of theNege v ismad e up by the Arava rift valley, running from the southern part of the toth e northern tip of theRe d Sea near Elat. TheArav a valley separates theNege v from the mountainous regions of Mcab andEdom , both ofwhic h arepar t today of the Hashemite Kingdom of . The border between the latter kingdom and theNege v was demarcated by theBritis h goverment in 1922, and follows the lowest course through the Arava rift valley. The northern boundary of the Negev isno t well defined. In the coastal and southern Judean foothills, it is situated ina semi-arid transition zone of dry farming.Her e the boundary coincides more or less with the present 350 mm rainfall isohyet, roughly marking the lower limit of wheat growing without the need ofauxiliar y irrigation inmos t years. In this area the boundary may be depicted as running from theMediterranea n Sea near Gaza to (ancient Ziqlag)a t the south-western corner of the Judean Hills (Orni and Efrat, 1971). While the boundary continues further eastward along the southern extremity ofth e Judean anticline,th e rainfall isohyets sweep to the north asth eNege v mergeswit h the Judean Desert. The area between Arad and Sedom at the southern tip of theDea d Sea forms the diffuse boundary between the twodeserts . The total area of theNegev , within itspresen t political and administra­ tive boundaries,encompasse s about 12,000 square kilometers.Rainwater-harves ­ ting agriculture or runoff farming waspractise d inantiquit y by a sedentary population in the central Negev hills (roughly 3000 years ago and in the period from about A.D. 100-700) in a region of some2,00 0squar e kilometers, ofwhic h only 2% was actually farming land (Kedar, 1967), made suitable for runoff agriculture in the course of time by man-made engineering works. This region, henceforth referred to as the runoff farming district, is the prime geographical object of this dissertation, together with the Kadesh Barnea oasis in Sinai, just across the border with Egypt.A sha s already been noted, there isn o clear geographical boundary between theNege v and Sinai. Thus, Kadesh-Barnea can be regarded as lying in an area which forms the natural

10 extension ofth ecentra l Negev hills into Egypt.I n thisperspective , it isn o surprise that the area of ancient runoff farming found in the Negev continues across theEgyptia n border into Sinai. The arealexten t and boundaries of the runoff farming regions in the Sinai have not been established yet.

GEOGRAPHY OF THE RUNOFF FARMING DISTRICT

The principal work to determine the extent and distribution of ancient runoff farming in the Negev was carried out by Kedar (1967). The runoff farming district inth e central Negev hills encompasses an area of about 2,000squar e kilometers (200,000hectare s or 2,000,000 dunams), which is nearly 17% of the entire Negev. The combined total area of all the agricultu­ ral fields within the runoff farming district has been some 4004 hectares. Thus only 2% of the entire district wasmad e up by cultivable fields. These ancient fields are situated in the valleys, where the runoff waters tend to concentrate naturally, after having been generated by a rain storm. Only in the valleys are the soils deep enough to function asa wate r storage medium and buffer for agricultural purposes. Runoff agriculture requires two principal landscape units: (1)a runoff producing territory, and (2)a related, lower lying, runoff receiving area, consisting of cultivable land. The ratio between runoff receiving area and runoff producing area depends upon the climate, but isals o dictated by the composition of the landscape. The larger the amount ofrainfall , the smaller the required size ofth e runoff contributing area.Th e relation between runoff producing hills and runoff receiving valleys determines the potential area suited for runoff farming.A good understanding of the landscape iscrucia l in the planning of runoff farms. The runoff receiving soils,usuall y situated in the valleys,mus t be suited for this type ofagriculture .Sufficien t depth and a good water-holding capacity are important attributes in this respect. The average ratio in the runoff farming district between runoff receiving fields and related runoff producing areas is 1:18. The range of variation of this ratio within thedistric t runs from 1:9 to 1:33 (Kedar, 1967). This isa nearly four-fold difference, which cannot be explained interm s ofclimat e as average climatic differences within the district are small. It demonstrates the effect of local landscape properties upon the ratio. The main varying factors are natural drainage density, the natural size ofth e valleys and the surface quality ofth e runoff contributing area in terms ofrunof f production.

11 FIGURE3 .Geograph y of the runoff farming district

12 The runoff farming districti sa hill y region, rangingi naltitud e from about 200meter si nit snorth-wester n partt o100 0m i nth e south. -Th e NORTHERN boundary ofth e district runs from wadi Azariq atth e Egyptian border, near theconfluenc eo fNaha lNizzan a andNaha lLavan , ina north-eastern direction alongth efirs t anticlinal rangeo f Har Qeren. The boundary follows the southern edgeo fth e Halutza towards the remains ofHalutz a (Elusa, Hallasa), oneo fsi x ancient Nabatean-Byzantine towns inth erunof f farming district. The other towns areNizzan a (Nessana, Auja Hafir), (Rehovot Banegev, Ruheiba), (Sobata, Subeita), (Oboda, Abdeh), andMamshi t (,Kurnub) .Th efirs t namefo reac h sitei sth eon euse do nmoder n Israeli maps, whereas alternative spellingso r names usedi nth eliteratur eo ro nolde r maps, accordingt oNege v (1982),ar e put behind brackets.Thes e towns were abandonedi nth ewak eo fth e Muslim-Arab conquest inth e7t hcentur y A.D. From Halutza, thenorther n district boundary continues eastward, alongth e southern edgeo fanothe r fieldo fsan d dunes, toNaha l Sekher (Wadi Mashash). Hereth eactua l boundary differs somewhat from theon e drawnb yKeda r (1967), as remainso frunof f farmingar e alsopresen t along the valleyo fNaha l Sekher and Nahal Mingar. Theare a around the modern towno fDimon aan dth e ancient town of Mamshit constitutesth enorth-easter n cornero fth e runoff farming district. -Th eEASTER N district boundary runs from herei na south-wester n direction along Nahal Mamshitt oth eMakhtes h Hagadol, animpressiv e cirque in the Hatira anticline. Fromth e north-western cliffo fth eMakhtes h Hagadol, the boundary sweeps westward through Sede Boqer, circumventingth e largeZi n Valley with itscanyons , while moving back eastwardt oth e Avdat area.Th e boundary continues southward along theNaha l Aqev clifft oHar * Aqev, subse­ quently bending eastt oth e edgeo fth e Zin Valleyan dagai ndu esout h along Har Gerafon towards the grandiose northern cliffo fth ever y large Ramon erosion cirque. Isolated remainso fancien t runoff farming fields are foundo nHa rArdon ,whic h forms the northeasternri mo fth eMakhtes h Ramon. -Th e SOUTHERN boundaryo fth e runoff farming district coincides with the impressive cliff, which constitutesa veritabl e environmental border-line. OnHa r Ramon,th edistric t reacheswit h 1025m it s highest alti­ tude.Fro m here the boundary moves slightly east around thewesternmos t exten­ siono fth eMakhtes h Ramon, before continuing south-westward alongth e cliff ofMezoq e Lozt oth eEgyptia n border. Kedar divided the runoff farming district into1 0zone s (Figure 4), and * Har= hill or 13 \(r~>^^ y^^^ i«rV• !>SS7!^-..cn .' ?): .«fey**^- - ..-•''•"'••.rr!—-^ .jf \ ~*=sr r^_^.v . :// \ \_. .—•"" • 1 •., ^--^ r:: ; ¥ HAL;UZ A vS\.."— <^X. \ ' J '; *\.

.J/ [ ''' :\ . * \--'""'• ^"^-Ox 'XZM'AMSHIT -X v^^ o ,/"-.. S \ - .R EV 1V 1 M .--^SJ J^J"\ ' •*> J j<7 REHOVOT ^^T'"""""'--. ."X" '""V"-<' '^;'' ..--—^\V 1 ')/ Jf s^—/^ :+iS !'•-.- ^\ ', \''""""l .-V--:.''*•"'•--- . /4?X '•'/ V^ e==^ /' S '^^^' Si '-Nr., ..>--•. iVEROHAMF \v

^F|L\>/ V; A ^T^^y \ A S\H ALIM ^-^ /' "^P V \\ "~\.. /' '•---''• / \*yt <*7 \ 'V-A' :r'\ M\^ .-" 1 (S EDE . \ BOQER"'"" 7 3*^" \ **\ •> A r~- '""-- \ \ '' *v-- \ ' * ~'Hs ^^ \ \-i\-\ \ x-.,'*\""\ "'•-\ X>~r\ """•;• \ /' y ~77r~^<^ \ •• ^,''1 \ '"*"' ''^' ,* 1' \ /^*-fljJ^o^rti','T' \ :\^\"-.V?., ~'-0 >'•.-., •'. V'v .,— -.A '•JtMjkl ^

\ \ • \^-.\v-".-'Y? \\ - -s-i; • /1\ N ..'->.^. •• . .•;' VJASs^.y^t ^•M»k*3 t& f \ \\>-\,,.- h \\ '• i .f \ ' !UO \'\v*'\\ V, \\ V" jy •••••Y,O -./If* \\ • ^i\'v- ,v. v~- \rSr^C'^O V v, •^S s~—^ ± \AVDA T '•"vjvJU\! M cI t W \ / •' i ;.x—\ 2/y \ '•-./ \ /'? •'\ N ^TSfc..,.^ii,_ ,,\ \ ) \ ">•. V^-v '*•••) ••*('""'• \ X""^" 1' ! _V) \ i i • "; '\>'-''-• "7/ (""' :' \ )''i / -V"--. /' V^-M\\o*: A --\ ^~-\-;'- /"\ / •••' \ a.- •-. v- • y >VV; ••:.-'x \ «: i , K _y((W\ ^K"i-'K .-A.T£5~v A\\\/ ""\ ''• ''' J^^ % N Jr^ \ ' i '•' ' V'""^Ni: > ^ \ VV v:V -•" 1^/^ W^\ 0 \ : \ /iAi:K ;u, vxV-'!; % f- ;"2p^ itjr %$X * V-/Vi^\ >: Ic "••: /'i>\ ,/\* A: \ i Lo z j\ >.c LJ/ \Js£ ^7>TTV

"' HO RS H A '/ f jin*^"^

x ) .'' \ ' JjR -y

_^ . BOUNDARY RUNOFF FARMING ZONES I '''/i ^<„-<-'-., \ ij if-. '^'"jf; -?)l(fnff ESCARPMENT j£& SAND AREAS

o ANCIENT SITE

"""* WADI

0 10 20 30 km

FIGURE4 .Zona ldivisio no fth erunof ffarmin g district (basedo n Kedar, 1967)

14 determined for each zone anumbe r ofrelevan t data, which arequote d here, albeit in a somewhat different order than inth e original Hebrew version (Kedar, 1967, chapter 8). Thete nzone sar earrange d from west toeast ,an d from north tosouth . Thefirs t zone isth eHalutz a -Rehovo t area, situated inth enorth-wester n corner ofth erunof f farming district.Thi s zone isboun d by sand dunes inth enorth , west andsouth , andb yNaha lBeso r inth e east. The next sixzone sar elyin g between theforme r zonean da line , made up by the Avdat Plateau, theZi nValle y andth eMakhtes h Hagadol erosion cirque. These zonesar eliste d ina sequenc e going from west toeast :Nizzan a zone, Ruth zone, Shivta zone, Sede Boqer -Ashali m zone, Yeroham -Revivi m zone,an dMamshi t zone.Th ethre e remaining southern zoneso fth edistric tar e likewise listed from west toeast : Loz- Horsh a zone, La'ana- Yete r zone, and Avdat zone.

TABLE 2.Zona l division ofth erunof f farming district (based onKedar ,1967 )

NAME OFZON E TOTAL AREA AGRICULT. FIELDS %O F FIELDS %O F (hectares) FIELDS(ha ) ZONAL AREA ALL FIELDS

1. -Rehovot 15,800 239 1.5 % 6.0 % 2. Nizzana 17,000 1489 8.7 % 37.2 % 3. Ruth 9,200 237 2.5 % 5.9 % 4. Shivta 18,800 495 2.7 % 12.4 % 5. SedeBoqer-Ashali m 25,500 303 1.3 % 7.6 % 6. Yeroham- 27,900 227 0.9 % 5.7 % 7. Mamshit 6,200 42 0.7 % 1.0 % 8. Loz-Horsha 42,300 200 0.5 % 5.0 % 9. La'ana-Yeter 22,400 95 0.4 % 2.4 % 10. Avdat 12,500 678 4.5 % 16.9 %

Kedar (1967)als o determined within each zoneth earea ldistributio n ofth e runoff farming fields inrelatio n towadis , drainage basins,o rothe r geogra­ phic points ofreference . These relevant data,firs t published inHebrew ,ar e worthwile tob equote d aswel l (Table 3), asthe y givemor e insight intoth e respective importance ofcertai n catchment areasan d ephemeral in ancient runoff agriculture.

15 TABLE 3. Arealdistributio n of runoff farming fieldswithi n each zoneo f the runoff farming district (after Kedar, 1967).

RUNOFF FARMING DISTRICT EPHEMERAL STREAM VALLEY RUNOFF FARMING ZONE OR OTHER REFERENCE FIELDS (hectare)

1. HALUZA -REHOVO T Zone: Nahal Besor 101 Nahal Rehovot 128 Internal valleys 10

2. NIZZANA Zone: Nahal 455 Nahal Nizzana (within this zone) 1034 3. RUTH Zone: Nahal Ruth and its tributaries 124 Nahal Raviv 112 Nahal Safun and Nahal Perach 1

4. SHIVTA Zone: Nahal Shezaf and its tributaries 66 Nahal Qarhadrainag e basin 118 Nahal Zetan drainage basin 51 Nahal Derorim drainage basin 44 NahalLava n 126 Southern tributaries Nahal Lavan 155

5. SEDE BOQER Zone:Naha l Haroa 119 Nahal Boqer 19 Nahal Baqara 52 Wadi el-Umzira 25 Wadi um-Terafa 9 Nahal Besor (within this zone) 78

6. YEROHAM -REVIVI M Zone: Surroundings of Revivim 60 Surroundings ofMashabb e Sade 58 Surroundings of Yeroham 109

7. MAMSHIT Zone: North-western part 30 South-eastern part (diversion system) 12 8.L0 Z -H0RSH A Zone: Northward drainage area Nahal Nizzana (within this area) 83 Nahal Aqrav, Sirpad, and Ayarim 31 Nahal Horsha 49 Southward and eastward drainage areas 13 Westward drainage area (Nahal Loz) 25

9.LA'AN A -YETE R Zone: Nahal La'ana (northern part) 45 Nahal La'ana (southern part) 11 Nahal Yeter and its tributaries 40

10. AVDAT Zone: Nahal Zin drainage basin 167 Nahal Avdat drainage basin 449 Nahal Divshon drainage basin 48 Plateau between N.Zin and N.Div/shon 13

16 The amount of runoff farming fields related to the lower part of Nahal Nizzana (1034ha ) isabsolutel y outstanding. The second largest areal extent of ancient agricultural fields is inNaha l Ezuz (455 ha), also in the Nizzana zone, followed by the drainage basin of Nahal Avdat (449 ha). These latter figures are well above the average of the remaining catchments and the three mentioned areas combined represent nearly 50% of all theancien t agricultural fields in the runoff farming district.

GEOLOGY AND GE0M0RPH0L0GY

TheNege v islocate d at the edge of the Arabo-Afican craton, whose Precamurian rocks are only exposed in the south near Elat.Durin g the Cratonic Stage, lasting for most of the Phanerozoic until theOligocene , transgressions issued from the former Tethys Ocean (Mediterranean Area)i n the NNW, con­ trolled by mild vertical tectonic oscillations. Several sedimentation cycles produced a typical platform sedimentary cover ofmatur e elastics and carbo­ nates that were deposited over flat or in shallow epeiric seas (Garfunkel, 1978). Rather accentuated earth movements occurred in theLat e Triassic and during theLat e Jurassic -Earlies t Cretaceous.A new phase of tectonic activity, the results ofwhic h are significant even today forth e runoff farming district, began in the Senonian Period ofth eLat e Cretaceous and continued into the Tertiary. It produced a series of faults and broad asymmetric folds, all having a northeast-southwest to east-west alignment. These structures,exten ­ ding from theEuphrate s River in , through northern Jordan, Israel,an d Sinai into Egypt, comprise the Syrian arc ofKrenke l (1924). The associated folded anticlines inth e central Negev rise gently on thenorth-wester n side and dip often sharply on the south-eastern side, markedly influencing the present day geomorphology of the region. The lastmajo r transgression occurred during theEocene , by which carbo­ nates were deposited over large portions of theplatfor m (Benjamini, 1979). The sea extended probably 1000k m inland from the present coastline, but its retreat and the onset of terrestrial conditions inth e area began in theLat e Eocene, followed by aperio d of extensive erosion that produced a rather flat landscape (Garfunkel and Horowitz, 1966). Relics ofth emiddl e Tertiary topo­ graphy in the Negev are still preserved today (Garfunkel, 1978). Eocene rocks are predominant inth e runoff farming district (Figure5) .

17 The originally continuous Arabo-African craton was rifted apart in the Late , a process that perhaps began in the .Thes e earth movements produced the present plate boundaries in Israel and nearby countries.Th e Dead Sea rift forms part of the boundary of the new Arabian plate (Bentor et al., 1965; Freund, 1965;Garfunkel , 1978). Strong vertical movements considerably changed the pre-rifting topography of the region.Th e land surface surrounding the Dead Sea became the lowest continental spot on earth, situated about 400 meter below the level ofth eMediterranea n Sea. A variety of sediments and (), several kilometers thick, accumulated in the subsiding rift valleys. The areas adjacent to the rift valleys were often strongly uplifted and appear today asveritabl e mountainous walls. Especially the regions of Moab and east ofth e Arava Valley, look impressive when viewed from the eastern Negev.Th eMediterranea n coastal plain and offshore region to the west subsided a few kilometers since the Neogene.Th e northwestern Negev is,there ­ fore, rather flat and covered with loessial (Yaalon and Dan,1974 ; Bruins, 1976;Bruin s and Yaalon, 1979)an d sandy deposits. The boundaries ofth e runoff farming district clearly relate to the south­ west-northeast trending fold belt. The first anticline ofHa r Qeren, bounding thecoasta l plain, marks the northwestern limit ofth edistrict . The eastern and southern district boundary follows theHatir a and Ramon anticlines, and the intervenient steepmargin s of the Zin Valley which makes amajo r westward dent into the central Negev highlands (Figures 2,3,4,5). The agricultural fields within the runoff farming district are situated in valleys of ephemeral streams of first or higher order. Quaternary loessial deposits have accumulated in these valleys and form an ideal parent material for cultivable soilswel l suited for rainwater-harvesting agriculture. The hills and mountains ofth e district are largely composed of carbonate rocks formed under marine conditions during theUppe r Cretaceous and Lower Tertiary. The geological formations making up these hills,whos e lithological properties are important factors in runoff production, can be divided into three main groups: (1)Th e Group, (2)th eMoun t Scopus Group, and (3)th e Avdat Group. Runoff farming can only be successful ifenoug h runoff water is generated in a certain catchment area under the prevailing rainfall regime.Geomorpholo - gy and surface lithology of the catchment are important factors in this res­ pect. An outline is, therefore,give n about the lithological characteristics and areal distribution of the three geological groupswithi n the runoff far-

18 ming district. 1) The JUDEA GROUP ischaracterize d by HARD and DOLOMITE rocks of Cenomanian and Turonian age (Arkin and Braun, 1965; Bartov and Steinitz, 1974). Its rocks cover approximately 20% of the runoff farming district and dominate the anticlinal hills in the northeastern part, bound to the west and south by Nahal Kevuda, Nahal Zipporim and the Zin Valley. Extensive of the Judea Group are found in the Shivta, Sede Boqer-Ashalim, Yeroham- Revivim, and Mamshit zones on the following anticlinal ranges: Givot Kevuda, Ketef Shivta, Har Nezer, Har Boqer, Ramat Boqer, Har Haluqim, Har Rahama, Rekhes Yeroham, Har Shahar, Har Zavoa, and theHatir a anticline with Har Zayyad (Figures 3,4,5). Inth e southern part of the district the Judea Group crops out in theAvda t and Loz -Horsh a zones along the entire northern cliff of theMakhtes h Ramon, aswel l as further west onHa r Ramon, Har Horsha,an d Har Harif. Two small, isolated outcrops are found on Har Sa'ad and Har Nafha in the Avdat Zone.

2) The MOUNT SCOPUS GROUP usually consists of SOFT CHALK and MARL of Senonian, Maastrichtian, and Paleocene age (Flexer, 1968; Bartov et al., 1972). Its rocks cover about 5% of the runoff farming district and generally appear on the outskirts ofth e areas occupied by the Judea Group. .Relatively extensive outcrops occur in the central part of the district, in the Ruth and Shivta zones,wes t ofShivt a and Nahal Kevuda,an d north ofHa rLavan . Another major area is situated in the northern part of theYeroha m -Revivi m zone, northeast of Mashabbe Sade, in between Nahal Misad andNaha l Sekher (Wadi Mashash). Smaller outcrops are found in the syncline ofYeroha m and Sede Boqer, as well as north ofSheluha t Kadesh Barnea andHa r Hamran in the Nizzana zone.

3) The AVDAT GROUP iscompose d ofEocen e LIMESTONE and CHALK with variable amounts ofcher t (Bentor and Vroman, 1963;Braun , 1967;Benjamini , 1979,1980 ; Benjamini and Zilberman, 1979;Zilberman , 1981). Its rocks cover about 65 %o f the runoff farming district, appearing almost everywhere except in the north­ eastern part ofth e district.

Neogene and Quaternary terrestrial deposits, composed ofcoars e elastics, sand and loess, make up the remaining 10-15% of the district. Neogene depo­ sits are found particularly in theDimona-Yeroham-Sed eBoqe r syncline.Quater ­ nary sediments arewidesprea d in the northwestern part of the district and

19 30 =J km

Legend: ^ZHZ JudeaGrou p (Limestone and dolomite) District boundary 1111I Moun t ScopusGrou p (Chalk and marl) Prominent cliffs • •'.Avda t Group (Limestone and chalk) Geolog. boundaries

FIGURE 5.Geolog y of the runoff farming district

occur also inal l theephemera l stream valleys. Asalread y mentioned,Keda r (1976)divide d the runoff farming district into 10 zonesan d determined foreac h zone the total areao fancien t agricultural

20 fields (Tables 2,3 ;Figur e 4). A comparison between the landscape factors and the percentage of ancient agricultural fields does show aclea r relationship in certain zones, asdiscusse d below. The influence of socio-economic and political factors,superimpose d upon the geo-environmental zonal potential,i s indicated by the density of runoff farms around urban centres.Th e three zones with the highest percentages of ancient fields (8.7% , 4.5 %, and 2.7 %) are those related to theNabatean-Byzantin e cities ofNizzana , Avdat,an d Shivta, respectively. However, the distinct influence of geomorphology and lithology upon the area that could be reclaimed in each zone by runof farming in antiquity seems, nevertheless,apparen t in anumbe r of cases.Whe n theNizzan a zone is compared with the Halutza -Rehovo t zone, it isclea r that both zones had urban centres and large areas ofpotentiall y cultivable land with loessial soils. The Halutza -Rehovo t zone, which contained larger towns in antiquity than the former area, has, nonetheless,onl y 1.5 % of runoff farming fields,compare d to 8.7 % for the Nizzana zone. The respective landscape characteristics ofth e two zones seem to be the principal causes for this difference. The geomorphic division between runoff producing hills and runoff receiving cultivable lands, aswel l as the large catchment and suitable terrace plains ofNaha l Nizzana, are clearly advanta­ geous for runoff farming in the Nizzana zone. There isa clea r lack of runoff producing hills and toomuc h flat land inth eHalutz a -Rehovo t zone, as well as in the Yeroham -Revivi m zone, due to the extensive valley plains of Nahal Besor and Nahal Revivim. Moreover, the widespread occurrence in and around the latter regions of sand dunes, which absorb all the rainwater without genera­ ting any runoff, isa n additional lithological disadvantage that also limits the runoff farming potential. The other zones are situated inmor e mountainous and rocky terrain. The ratio and division between runoff producing hills and runoff receiving valleys seem to be themai n geomorphic factors that determine the potential area suited for runoff farming. However, the properties ofth e individual hill- slopes and valleys, in terms of runoff production and storage, respectively, are also very important. The local effect of lithology,surfac e covering,an d slope angle in runoff production was investigated indetai l in two different case studies. Shanan and his colleagues (Evenari et al., 1971;Shanan , 1975;Shana n and Schick, 1980)stres s the runoff producing qualities of gentle, loess covered slopes, free of stones,whic h they investigated in theEocen e Avdat area.Yai r

21 (1981, 1983) arrived at quite a different conclusion inhi s studies of an instrumented hillslope section on hard (Turonian) limestone rocks of the Judea Group in the SedeBoqer-Ashali m zone. Yair considers outcrops and steep stony slopes asadvantageou s for theproductio n and transmission of runoff.Gentl e loesscovere d slopes play a negative role inhi s opinion. These seemingly contradictary view-points appear both to be correct within their own context. More comparative research isneede d toevaluat e the ramifications of these different results interm s of various runoff farming systems (Chapter 3),e.g . micro-catchments on very gentle sloping loess versus other systems that receive their runoff water particularly from rocky slopes. The hillslope runoff model ofYai r yet requires, as he acknowledges (Yair, 1983), anappro ­ priate explanation for the function, in hismodel ,o f the already much debated stonemound s on thehillslope s (Tadmor et al., 1958;Shanan , 1975).

CLIMATE

The Negev is situated in the northern, subtropical part, of the large planetary desert belt that extends from to India. Aridity in the Negev is,therefore ,mainl y the result ofdescendin g dry and stable air masses ofgloba l high-pressure cells, situated beyond the tropical convection zone. Climatic conditions are typical of awinte r rainfall desert.Th eplane ­ tary pressure system shifts southward in winter, bringing Israel under the influence of awesterl y circulation system with rain bearing depressions from theMediterranean . TheNege v is situated on the southern edge of this circula­ tion system inwinter , asmos t depressions follow amor e northward trajectory over Cyprus. Mean annual precipitation decreases,a s aresult , from about 300 mm in the northwestern Negev to some 25m m in the extreme south near Elat. Depressions sometimes arrive from the south along anarro w belt over the and cause sudden cloudbursts accompanied by torrential floods in the Negev and Sinai deserts (Ashbel, 1973). The average temperature in Januari, the coldest month,i sabou t 12 degrees Celsius in the north-western Negev and nearly 16degree s inElat . The hottest month isAugust ,havin g average temperatures of about 27 in the north-west and 33 degrees Celsius in the extreme south. Daily maximum temperatures in summer are regularly above 40degree s Celsius in the Arava rift valley (Rosenan, 1970). P/ETP indices range from0.39 0nea r Gaza in thenorth-wester n Negev to

22 600 0.009nea r Elati nth e extreme south. Aridity, therefore, ranges from semi-arid inth e north­ 500 west tohyper-ari di nth e southan d east, as definedb yUnesc o (1979). The climateo fth e runoff farming district in the central Negev hillsi stypicall y arid, situated betweenth esemi-ari d northern Negev and the hyper-arid regionst oth e south and east.Th eari d region actually protrudes south­ ward intoth e hyper-arid surroundings, duet o the elevationo fth e central Negev. However, only the rising northern and western parts of the central Negev highlands, directed towards the trajectory ofrai n bringing depressionsi n

Sede Boqeri;-"' winter, receive relatively more rain.Th e eastern and southern partso f the highlands receive less rain,a ssoo na sth e general topo­ > \ \ graphy beginsa downwar d trend, becauseo fth e rainshadow effect and increasing structural aridity southward.I ti squit e striking howth e eastern and southern boundaryo f the runoff Elat farming district coincide with topographic features, which mark amor e or less sharp FIG. 6. Rainfall isohyets decline inelevation . This surely isa major climatican denvironmenta l border. Average annual rainfall amounts inth e runoff farming district range from about 120m mi nth enort ht osom e7 5m mi n the south-east. These dataar e basedo nvariou smeasurement si nth e area in the courseo fth e 20th century,a sshow ni nTabl e 4.Mos to fth e precipitation occurs erratically between Novemberan dApril . Interannual rainfall variabili­ tyi shigh ,bein go nth eaverag e above5 0 %. The rainfall regimei nth e central Negevi susuall y very gentle with low rainfall intensities (Katznelson, 1959;Shana ne tal. , 1967). Theaverag e amount ofrain y daysa tSed e Boqer, situatedi nth ecentral-easter n part of the runoff farming district,i s26 ,base d upona 2 8yea r record from 1951/52- 1979/80. The mean annual rainfall over thisperio di s93. 5mm . About halfo f these rainy days register less than1 m mpe r day.Mor e than1 0m mrain/da yca n be expectedo nonl y three dayspe ryear , onth e average.

23 TABLE4 .Averag e annual precipitation inth erunof f farming district.

Recording Location Elevation RunoffFarmin g MeanAnnua l Period (m) Zone Rainfall (mm)

1943-1963 Revivim 290 Yeroham-Revivim 101 1934-1962 MashabbeSad e 350 Yeroham-Revivim 103 1970-1984 WadiMashas h 370 Yeroham-Revivim 117 1942-1948 Mamshit 450 Mamshit 135 1901-1930 Nizzana 250 Nizzana 86 1960-1976 Shivta 350 Shivta 88 1951-1980 SedeBoqe r 470 SedeBoqer-Ashali m 94 1960-1984 Avdat 565 Avdat 87 1958-1978 MizpeRamo n 890 Loz-Horsha/ Avda t 78

A rainfall of2 5 mm/day isexpecte d nomor e than once every twoyear san d a rainfall of5 0mm/da y ishighl y improbable (Shanane tal. , 1967).A nanaly ­ sis of rainfall intensities atAvdat , during theperio d 1959/60 - 1971/72, showed that main rain-storms, important forrunof f farming, canb e divided into three principal types (Shanan, 1975):

- Type A consists ofa principa l rainfall lasting forabou t 6-8 hours, during which period some 10m mo frai n falls. Themaximu m intensities willb e relatively low: 6-7mm/hou r fora perio d ofa quarte r ofa n hour, and 3-4 mm/hour fora on ehou r duration. After twohour sth eaverag e storm intensity dropst obelo w2. 5mm/hou r andlevel soff a tabou t 1.5mm/hour . This type of rain-storm occurs virtually every year.

- TypeB consist so fa principa l rainfall lasting forabou t 6hours , during which period some1 2m m falls.Maximu m intensities forshor t durations are 60- 100 % greater than forth eforme r typeA . Intensities of12. 4 mm/hour may occur during aperio d ofa quarte r ofa nhour , and5. 3mm/hou r fora on ehou r duration. After2 hour sth eaverag e rainfall intensity is3. 2mm/hour , drop­ ping to2. 4mm/hou r only after some4 hour so frain .Th eaverag e intensityfo r thesi xhou r rain-storm period will be2 mm/hour . Thistyp eo frain-stor mca n be expected tooccu r once every twoyears .

24 - Type C consists of asingl e rainfall of extremely high intensity, which occurred on only one occasion (5-10-1965) in the period from 1959/60 - 1971/72. The entire rain-storm produced 35m m of rain in six hours.Thus ,th e average rainfall intensity was almost 6mm/hour .Maximu m intensities for short periods: 28mm/hou r for aperio d of aquarte r of anhou r or half an hour, 19 mm/hour for aon e hour duration, and 13mm/hou r for a2 hou r duration. This type of rain-storm with aduratio n of 4 hours seems tohav e a frequency of once in five years, and with aduratio n of6 hours once in about 20 years (Shanan, 1975).

Temperatures in the runoff farming district are ofcours e influenced by altitude. Mean annual temperatures range from about 20degree s Celsius in the northwest ofth e district (200-350m )t o 17.8C atMizpe h Ramon (890m ) in the south. Mean temperatures of the coldest month (January)an d hottest month (August)ar e about 12C and 27C , respectively, in the northwest and about 9C and 25 C in the south. Differences between day and night temperatures are typically large for an arid desert region,bein g about 14degree s Celsius on a yearly average (Rosenan, 1970). Frost may occur in the valleys of the runoff farming district during 20 to 30night s inwinter , especially in the more elevated parts of the district (Evenari et al., 1971). During Hamsin'or Sharav weather conditions in spring and autumn, hot and dry winds from the Arabian desert may raise the temperature to4 0 degrees Celsius ormore . Dew is an important phenomena in the central Negev hills, occurring on approximately 150-200night s each year. The average annual surface dewfall at Avdat isabou t 33m m (Evenari et al., 1971). Although the effect of dew on cultivated crops has not been definitely ascertained, Duvdevani (1953) concluded from his experiments that it has agrowt h promoting effect.

SOILS AND VEGETATION

Soils occupy the uppermost part ofth e earth's crust, being in direct contact with the atmosphere. Their properties differ from the underly­ ing rock material, as aresul t of interactions between this rock material and the environment over periods oftime . Apart from of the parent rock, soils in theNege v are usually influenced by the influx of aeolian and (Yaalon, 1964;Yaalo n and Ganor, 1973). In thecas e of loess deposits the exotic dust from the air has become aparen t material of its own.Climate ,

25 living organisms, and landscape position are major soil forming factors. In Israel the landscape approach was adopted in the study of soils, empha­ sizing theeffec t ofgeomorphi cpositio n and parent material on soil develop­ ment.A n impression ofth e areal coverage ofparen t materials in the Negev has been summarized by Yaalon (1981). About 60t o 65% of the area iscompose d of rocky deserts with bare rocks and desert lithosols. Some 15 %ar e regs of the sedimentary plains or covered by desert pavements. Another 5t o 10% are loessial plains with soils characterized by secondary carbonate formation at various depths. Some 10% are sand fieldswit h mostly shifting sand dunes, and about A% istake n up by ephemeral stream channels and alluvial fans. Since the vegetation of aregio n is also influenced by the environment, there usually exists astron g relationship between the soils of an area and its plant geography. Four phytogeographical regions ofcharacteristi c floral composition and obvious environmental significance have been recognized in the Negev:

1) The MEDITERRANEAN phytogeographical region is limited to the northern Negev fringe,bounde d by the 300m m rainfall isohyet. 2) The IRANO-TURANIAN region is situated between the 300 and 80 mm iso- hyets and its floral composition, therefore, dominates the runoff farming district. 3) The SAHARO-ARABIAN region generally coincides with the hyper-arid zone, limited to areas receiving less than 80m m ofmea n annual rainfall. 4) TROPICAL SUDANIAN flora has infiltrated intoth eho t Arava rift valley, bounded by the isotherm of 23 degrees Celsius (Zohary, 1966; Danin, 1983).

Vegetation inth ehyper-ari d Saharo-Arabian region isrestricte d to favou­ rable likeephemera l stream channels and rock crevices, where the runoff water concentrates naturally. Both fluvial and wind erosion issevere , due to the absence ofa vegetationa l cover. The surface deposits are, there­ fore, depleted of fine soil material,whic h ispartl y redeposited as loess in the semi-arid desert fringe (Yaalon and Dan, 1974). As weathering is slow in the very dry climate and erosion severe, soils are not well developed. Desert Lithosols (Lithic Torriorthents) and Coarse Desert Alluvium (Typic Torriflu- vents) are characteristic soils of the hyper-arid region. Stable landscape surfaces like sedimentary terraces and plateaus, covered and protected by (Reg or Hammada), exhibit the best developed authigenic desert

26 soils (Dane t al., 1982), called Reg soils in Israel (Gypsiorthids and Cambor- thids in the American soil taxonomy system; USDA, 1975). North ofth e runoff farming district, inth e semi-arid part ofth e Irano- Turanian and Mediterranean floral regions, the amount of rainfall is usually sufficient to enable the formation of a rather continuous vegetational carpet. Soil erosion isrestricted . Aerosolic dust from theSina i and deserts could, therefore, accumulate to form rather thick and continuous loess depo­ sits in the north-western Negev (Yaalon and Dan, 1974; Ganor, 1975;Bruins , 1976; Bruins and Yaalon, 1979). Soils arematur e with well developed soil profiles, exhibiting calcic horizons (Light Brown soils; Haploxeralfs,Pale - xeralfs,an d Xerochrepts). The arid runoff farming district ischaracterize d by adiffus e vegetational cover, dominated by an Irano-Turanian floral composition.Th e scattered vege­ tation cannot prevent severe erosion. However, in favourable habitats like valleys,wher e the runoff concentrates,vegetatio n isofte n sufficiently dense to enable deposition of loessial fines eroded from thehillslopes . Soil deve­ lopment is usually slow, but somewhat more marked than inth e hyper-arid zone (Dan, 1981). The eastern and southern boundary of the runoff farming district is ofmajo r environmental significance: This line coincides with the south­ eastern boundary ofbot h the soil association ofBrow n Lithosols - Loessial Serozems (Figure 7)an d the Irano-Turanian floral region. The soil associations found in the runoff farming district are shown in Figure 7an d listed inTabl e 5. An estimation is given inTabl e 5 of their respective areal coverage inth e district, asdeduce d from the soilma p of Israel (Dan et al.,1975) .

BROWN LITHOSOLS ONHILLSLOPE S OF HARD CALCAREOUS ROCKS Rocky hills and mountains dominate the runoff farming district, constitu­ ting about 70% of the region. Weathering of thehar d limestone and dolomite rocks isver y slow,du e to the dry climate.Th e soils on these hard calcareous hills are, as a result, mainly formed from aeolian loessial dust mixed with parent rock fragments. These shallow soils, designated as Brown Lithosols (Lithic Torriorthents), are largely confined topocket s and crevices among the hillslope bedrock, where they are protected from accelerated erosion. Most of these soilsar e saline, but the upper soil layer has a low salt content due to the leaching effect of the rainfall.Salinit y in the subsoil is less in places where runoff from surrounding bedrock outcrops can accumulate. The additional moisture from runoff water also enables the development ofa

27 BOUNDARY DISTRICT

ESCARPMENT 10 20 30

FIGURE7 & TABLE 5. Soil associationso fth eRunof fFarmin g District

SOIL ASSOCIATION GEOMORPHICPOSITIO N AREA

1)Brow n Lithosolsan dLoessia l Hillslopesan dvalley s 65 % Serozems 2)Loessia l Serozems Plains,plateaus ,valley s 15 % 3)Sand y Regosolsan d Loessial Serozems Nw plainsan dvalley s 10% A)San d Dunes NW plainsan dvalley s 5 % 5)Calcareou s Desert Lithosols Hillslopes 3 % 6)Re gsoil san dCoars e Desert Alluvium NahalNizzan a valley 1 %

28 relatively rich vegetation, mainly composed ofsmal ldeser t shrubs (Danin et al., 1975; Yair and Danin, 1980; Dan, 1981). The Artemisia herba alba - Reamuria Negevensis plant association covers most of the limestone hillslopes on the Avdat Plateau. This association is replaced by theArtemisi a herba alba -Gymnocarpo s decander,Artemisi a herba alba- Salvi a lanigera,an d theArtemisi a herba alba - Thymelaea hirsuta plant associations on the (Judea Group) limestone and dolomite hillslopes ofth e northern part of the runoff farming district. The various Artemisia associations are replaced inth e drier parts of the district and on south facing slopes by the Zygophyllum dumosum -Reamuri a negevensis plant association (Danin, 1983).

CALCAREOUS DESERT LITHOSOLS ON HILLSLOPES OF SOFT CALCAREOUS ROCKS Weathering of softcalcareou s rocks inth e runoff farming district, like chalk and marl,i ssomewha t faster.Soil s developed onhill s composed of these parent materials areno t only formed from exotic airborne loess. They contain a proportionally larger share ofweathere d rock fragments than soils developed on hard rocks. Soilso n chalk and marl,therefore ,reflec t to alarge r extent the composition of the underlying parent material. These soils are included among the Calcareous Desert Lithosols (Lithic Torriorthents)an d cover certain parts of the Shivta,Ruth ,Nizzana ,an d La'ana-Yeter zones. Calcareous DesertLithosol s are more saline thanBrow nLithosol s on hard rocks. Outcrops of bedrock are less common and vegetation is scarce (Dan, 1981). Reaumurietum negevensis, Reamurietum hirtellae, andman y other plant associations, often related to certain geological formations, are found on these soft calcareous rocks (Danin, 1983).

YOUNG ALLUVIAL SOILS INTH E ACTIVE STREAM CHANNELS The valleys ofth erunof f farming district are largely covered by loessial sediments. Coarse gravel deposits are generally confined to themajo r , like the broad Nizzana Valley. The younger sediments associated with active epehemeral stream channels do not show soil profile differentiation. Any textural variations are essentially due to fluviatile sedimentation. These young soils arecalle d Alluvial Loess (TypicXerofluvents )an d Coarse Desert Alluvium (Typic Torrifluvents), both of which are largely non-saline. Carbonate content in the Alluvial Loess soils varies according to rock charac­ teristics in the surrounding area.Thes e soils are highly calcareous incatch ­ ments dominated by chalk and marl, when their vegetation is often characte-

29 rized by theAtriple x halimus- Achille a fragrantIssima plant association (Danin, 1983). Carbonate content inAlluvia lLoes ssoil s ismuc h lower in areaso fhar d limestone and dolomite rocks (Yaalon andDan , 1974).

The NON-SALINE ALLUVIAL LOESS SOILS situated inactiv e ephemeral stream channels, represent the BEST WATERED inth erunof f farming district.Thi shabita t benefits from runoff floodsever y year. Thevegetatio n isofte na dens e of shrubs, such as Thymelaea hirsuta and Retama raetam, plants like Achillea fragrantissima, and various annual and perennial grasses, as well asherb sbelongin g to the more humid Mediterranean floral community. The latter includeAven a sterilis,Oryzop - sis miliacea, Hordeumsponta - neum, Lolium rigidum, Cynodon dactylon, Trigonella arabica, PHOTO 1. Spring vegetation in theWester n Medicago spp, and others wadi ofHorva tHaluqim ,jus t upstream from (Hillel and Tadmor, 1962; thecister n and terraced fields. Danin, 1983).

A most remarkable feature ofth e runoff farming district isth e occurrence oflarg ePistaci a atlantica treesa t higher elevations.Th ehighes t density of trees is found above 700meter . Altitude is, however,no tth emos t important factor in thedistributio n ofth e atlantic pistachio, asmos t of these trees appear inth eLoz-Horsh a zone in association with largeoutcrop s of smooth­ faced limestone,idea l inrunof fproductio n (Danin, 1983).

LOESSIAL SEROZEMS Older Loessial depositso n naturalwad iterrace san d in synclinal plains exhibit a distinct profile differentiation. A shallow calcic horizon and

30 saline layers at depth characterize these soils, defined as SEROZEMS. is usually present and some clay illuviation may alsooccur . Salinity in the subsoil reaches 2% ofsolubl e salts or even more. The upper 30 cm are almost free of salts,markin g the average depth of effective rainfall penetration. There isofte n agoo d correlation between salt content of the soils and the stability and age ofth epedomorphi c surfaces.Thi s indicates that most of the salts are airborne, slowly accumulating with time in adr y non-leaching envi­ ronment (Dan and Yaalon, 1982). The Bhorizo n of the somewhat less developed Loessial Serozems iso f acambi c nature and hasa loamy texture (Camborthids or Calciorthids). Morematur e Serozems, usually on the higher terraces, may display anargilli c Bhorizo n ofcla y loam texture (Haplargids). Petrocalcic stony Serozems on Neogene or Early Pleistocene geomorphic surfaces exhibit strongly developed calcic horizons and even calcrete (Dan, 1981). Serozems are found in all the zones of the runoff farming district,especi ­ ally in the synclinal plains of the southwest-northeast trending fold belt and in the valleys ofNaha lNizzana , NahalKevud a and Nahal Lavan. They also occur, however, inPleistocen e aeolian loess on high plateaus in the Avdat and Loz-Horsha zones of the central Negev mountains. The vegetation on the Serozems isa t lower altitude characterized by scattered low shrubs of Hamma- detum scopariae and Zygophyllum dumosum, and above 700mete r by Anabasetum syriacae (Danin, 1983). The present moisture regime isgenerall y poor,a s most loessial plains, plateaus and natural wadi terraces dono t receive runoff water. It seems that most Serozems are paleosols formed in a somewhat wetter arid to semi-arid climate during theLat e Pleistocene (Dan et al., 1973;Issa r and Bruins, 1983).

SANDY SOILS Sand occurs extensively in the northwestern part ofth e runoff farming district. Most ofth e young aeolian sand dune deposits dono t reveal any soil development (Torripsammentsan d Quartzipsamments). The beginning ofpedogene ­ sisma y be detected inth e young sandy Regosols (Typic Xerorthents) developed on stable sand covers.Som e darkening of the upper soil layer has taken place, aswel l as incipient lime segregation inth e subsoil.Th e soils underneath the sand on the foot-slopes of hilly areasmainl y originate from chalks and con­ glomerates. These Brown Lithosols and Rendzinic Desert Lithosols (Torri- orthents)ar e mixed with sand in their upper soil layer. Soils underneath the sand in the valleys and plains include various Serozems, aswel l as young alluvial loess deposits (Dan, 1981).

31 The vegetation of the sandy areas isquit e lush, in spite of the dry climate.Thi s isdu e toa favourable moisture regime asa result of deep rain­ fall penetration into the sand. The dominant plant of themobil e sand dunes, and sometimes the only species present, is the perennial grass Stipagrostis scoparia. It grows only in sites that are continuously being covered by new sand. Vegetation ofth e stable sand fieldsconsist s ofa mixe d community of shrubs and grasses, dominated by theArtemisi a monosperma -Retam a raetam plant association, accompanied by Thymalea hirsuta, Convolvuletum lanati, Echiochilon fruticosum,Neurad a procumbens,Iri smaria egeophytes ,an d grasses like Stipagrostis ciliata, Stipagrostis obtusa, Panicum turgidum,Pennisetu m divisum, and Aristida species (Hillel and Tadmor, 1962;Waise l et al.,1978 ; Danin, 1983). Inmor e humid regions of theworl d sand constitutes the driest ofhabitats , due to rapid drainage and failure to retain sufficient moisture and nutrients. However, the opposite situation generally prevails in arid regions, where sandy soils offer more favourable moisture conditions forplan t growth than do finer-textured soils. This is because of the higher permeability to rain and, hence, greatly reduced runoff losses.Rainfal l isusuall y so scant inth e desert,tha t no appreciable percolation takesplac e beyond the reach of plant roots.Moreover ,sinc e sand retains lessmoistur e per unit of depth than finer-textured loessial soils, lessmoistur e remains near the soil surface, where it might be subject to subsequent evaporation (Hillel,1982 ,p.103) . Notwithstanding these assets, the virtual absence of runoff in sandy areas renders them unsuitable for runoff farming.An y form of agriculture, including rainwater-harvesting agriculture, requires much more water (250-500 mm)tha n the average annual precipitation (100mm ) in the area.Fo r the requirements of runoff farming, therefore, by which relatively large quantities of runoff water must be stored inth e root zone, the disadvantages of sand felt in more humid regions becomeapparent . Loessial soils, on theothe r hand, virtually omnipresent in thewadi s of the runoff farming district, arewel l suited for this ingenious and yet ancient form of desert agriculture.

32 3 Runoff farming in the region

INTRODUCTION

The crucial factor inth esettlemen t ofa n arid desert is the assured supply ofwate r forman ,hi sanimal san dcrops .Sinc eth eNege v isno t endowed with a river that bringsi nwate r frommor e humid regions, local rainfallan drunof fha sbee nth eonl y water source formos to fth epast . Only inth e20t h century itbecam e possible, after theestablishmen t of'th e State of Israel,t obuil ta pipelin e that carries water fromth emor e humid northern parto fth ecountr y toth eNegev .Th etransportatio n ofthi swate r isbecomin g increasingly more expensive, whereasth eamoun t thatca nb eallocate d toth e Negev islimited . In197 5abou t 125millio n cubic meterso fwate rwa s piped intoth eNege v forirrigatio n agriculture,whic h is11. 5% o fth etota l amount ofwate r used foragricultur e inIsrae l (Hillel, 1982). Springs and wellsi nshallo w groundwater reservoirs, naturally fed by runoff waters, were theonl y available, more permanent, water sources in antiquity. The numbero fthes e shallow groundwater sourcesar efe w and the amount of water that canb etappe d from these springsan d is very scanty. Their water issometime s slightly brackishan dthei r potential for irrigation agriculture hasalway s been negligible, except forth e site of Kadesh-Barnea. Deep 'fossil' groundwater isknow nt oexis t inth eNege v at several hundredst omor e than thousand meter belowth esurface . Useo f this generally brackish water inantiquit y would have been unthinkable, as its possible utilization iseve n problematic today. Since springsan dwell si nth erunof f farming districtar efe wi n number, alternative solutions hadt ob emad ei nth epas t to ensure a satisfactory

33 infrastructure of potable water. Hundreds of cisterns were constructed in antiquity to collect and store some of the runoff rainwaters as essential forma n and his animals. Open-pit cisterns, known from the Israelite Period, are usually considered tob e older than roofed cisterns excavated into bedrock. Most cisterns of the latter type were probably con­ structed in theNabatea n and Byzantine Periods (Evenari et al., 1971;Hillel , 1982). Without cisterns the establishment of trade routes, towns and agricultural settlements would have been impossible in the past. The extensive form of livestock grazing in thisari d region isequall y impossible without a regular occurrence ofwaterin g points, which only cisterns could provide for. Runoff farming and livestock grazing, based upon local rainfall and runoff waters, have been the dominant types of land-use in the district during historical times.

TERMINOLOGY OF RAINWATER-HARVESTING AGRICULTURE OR RUNOFF FARMING

Before dwelling upon the essence of runoff farming, it seemsimpos ­ sible to bypass semantics, as ideas and logic are conveyed through words. In view of the variant usage ofword s in the literature related to the topic under consideration, aswel l asth e variousmeaning s adhered to these words, it seems clarifying tostat e at least the author'sposition ,alread y published elsewhere (Bruins et al.,1986a) . The term "water harvesting" seemswel l established inth e literature since itwa s probably first used by Geddes (1963), who defined it as "the collection and storage of any farm waters, either runoff or creek flow, for irrigation use". Myers (1975) modified this definition to "the practice of collecting water from an area treated to increase runoff from rainfall and snowmelt" (cf. Hollick, 19S2). Whatever definition ispreferred ,wate r harvesting clearly has ahydrologi - calmeanin g and cannot be regarded as an agronomic term.Moreover ,o n the face of it,wate r harvesting doesno t specify what kind ofwate r isharvested , i.e. rainwater,river-water ,groundwater ,etc. ,neithe r forwha t purpose itwil l be used. Water harvesting iso f course a term in itsow n right,bu t itcanno t be an alternative for runoff farming, since the latter term is intrinsically agronomic incharacter . "Rainwater harvesting" (Boers and Ben-Asher, 1982)i salread y more specific

34 as far as the source of the water isconcerned . Yet,als o this term ispurel y hydrological in itswording . "Waterspreading" (Newman, 1963;Cunningham , 1975) and "waterponding" (Newman, 1966;Cunningham , 1970)ar e hydrologic terms used in to describe the control of runoff water from rain-storms. "Floodwater farming" (Bryan,1929 )i sa fundamentally different term, as it clearly states thepurpos e for which the floodwater is used: farming. The question iswhethe r thishydro-agronomi c term can beregarde d as fully synony­ mous with runoff farming. From ageomorphi c point of view, floodwater is derived from a stream channel, either occupied by an ephemeral.stream or by a perennial river.Example s of floodwater farming from ephemeral streams are the Ak-Chin ( mouth) fields of the Indians ofArizon a (Hack, 1942)an d the Limans at the wadiMashas h experimental runoff farm in the Negev desert (Evenari et al., 1982). The early agricultural systems on the floodplain of the Nile (Butzer, 1976)ar e a form of floodwater farming related to aperen ­ nial river. Although nearly all thewate r used for agricultural purposes is somehow derived from rainwater, it seems better to make adistinctio n between epheme­ ral streams and perennial rivers, as far as irrigation farming is concerned. The relaxation time between actual rainfall and the eventual use of this water for agriculture isusuall y short in the case of floodwater farming from ephe­ meral streams, but very long forperennia l rivers.Sinc e "floodwater farming" may seem to include both of these quite different situations, the term isno t specific enough to characterize the type of agriculture under discussion. Moreover, the term excludes rainwater-harvesting agricultural systems that utilize runoff water collected as overland or sheet-flow, which cannot be classified as floodwater. Various micro-catchment systems (Evenari et al., 1971; Shanan and Tadmor, 1976; Boers and Ben-Asher, 1982), therefore,woul d not fit the term floodwater farming. There is, however, a need for a concise unifying term that is general enough to include all types of agricultural systems based upon rainwater harvesting methods.I norde r to harvest rainwater from whatever type ofcatch ­ ment, the rain has to flow asRUNOF F over a surface or ina channe l before it can be stored into the soil, a cistern,a pond,a ,etc .Fro m the storage medium, the water can then be made available, passively or actively, for agricultural purposes. Hence the term RUNOFF FARMING or RUNOFF AGRICULTURE seems precise, short and yet general enough to function as a comprehensive hydro-agronomic tag to label the agricultural systems under discussion. As such, the term has been used by Evenari and his collaborators during the last

35 25 years (Evenari et al., 1982). As already mentioned previously, Huibers (1985)ha scorrectl y drawn atten­ tion to the fundamental difference in the use of runoff rainwater for farming in semi-arid or arid regions.I nth e semi-arid zone direct rainfall isnormal ­ ly sufficient to sustain a crop. Runoff water isher emerel y collected as a by-product from the local rainfall, useful to irrigate the crops intime s of moisture stress. However, in the arid zone runoff water isabsolutel y essen­ tial to practise agriculture based upon local rain, since direct rainfall alone ishardl y ever sufficient. Rainfed farming inth e semi-arid zone with additional runoff collection (Huibers, 1985)is ,therefore ,no t reckoned to be runoff farming as defined below. The hydrological term runoff isno t always properly understood by the average individual,sinc e runoff isno t a common word occurring in everybody's vocabulary. For that reason RAINWATER-HARVESTING AGRICULTURE may be analter ­ native general term instead of runoff agriculture,bein g less concise but more descriptive and perhaps easier understood. In this dissertion the terms runoff farming and rainwater-harvesting agriculture are used interchangeably and defined as follows: -RAINWATE R HARVESTING AGRICULTURE ISFARMIN G IN ARID REGIONS BY MEANS OFRUNOF F RAINWATER FROM WHATEVER TYPE OF CATCHMENT OR EPHEMERAL STREAM - (Bruins et al.,1986a,b) .

HISTORY OFRUNOF FFARMIN G INTH E NEGEV.

CHALCOLITHIC There are indications that already during theChalcolithi c Age,mor e than 5000 years ago, runoff farming was practised along theephemera l stream ofNaha l (wadi)Beershev a in theNorther n Negev (Levy, 1983). There seems as yet no evidence that check-damswer e used at that time. For a successful functioning of thesystem , the geomorphic position of the loessial floodplain during Chalcolithic timesmus t have been only slightly higher than the epheme­ ral stream channel,s otha t seasonal floodwaters could inundate the floodplain readily and wet the soil to sufficient depth. The present geomorphic situation israthe r different,a s the stream channel ofNaha l Beersheva and many other wadis in the area are in adowncy t position, well below the former floodplain level. Today,thi s particular type of runoff farming would not be possible at the same sites due to achang e in valley morphology.

36 BRONZE AGE The Central Negev Highlands were settled quite extensively during the Middle Bronze Iperio d (ca. 2200 -200 0 B.C.), and there is circumstantial evidence that some kind of runoff farming wasprobabl y practised in suitable wadis (Evenari et al., 1958;Cohe n and Dever, 1980).

ISRAELITE PERIOD Archaeological structures and terraced wadis, possibly contemporaneous with the former, might indicate that rainwater-harvesting agriculture was practised in the Negev during the Israelite IIperio d or Middle Iron Age (Aharoni et al.,1960 ;Cohen , 1976). These runoff farming settlements date,i n the opinion of Aharoni et al (1960)an d Cohen (1976, 1980), to the 10th century B.C., and are presumed to have been established during the reign of King Solomon. There is,however ,a clea r lack in chrono-stratigraphic data in theNege v concerning the end of theLat e Bronze Age and the first centuries of the Iron Age. Remnants of runoff farming sites that belong to the Iron age have also been found inth e Judean desert (Stager, 1976).

LATE NABATEAN-ROMAN-BYZANTINE-EARLY ARAB PERIOD After aga p of perhaps amillennium , runoff farming in the central Negev was again practised in the 1st century A.D. by theNabateans . In the opinion of Negev (1982a)i twa sprobabl y with the accession toth e throne ofRabe l II in A.D. 70, that the bulk of the Nabatean population abandoned its tents, exchanging them for excellently-built houses. From the late80 s to the late 90s of the 1st century A.D. the completion of runoff farmswa s inaugurated in the region ofAvda t (Negev, 1982a). During the Byzantine period, from the 4th-7th century A.D., rainwater- harvesting agriculture came to the peak of its development in the Negev desert (Mayerson, 1955,1960a ;Kedar , 1957,1967 ;Evenar i et al.,1971 ;Kloner ,1975 ; Negev, 1982b). The unique Nizzana documents from the 6th and 7th centuries A.D., discovered in 1937b y the Colt expedition in two ancient Byzantine churches during excavations in Nizzana, mention the growing ofwheat ,barley , aracus (a legume), grapes,olive s and dates.Ther e isals o evidence that figs and were cultivated (Kraemer, 1958;Mayerson , 1960a). TheMuslim-Ara b conquest in63 6 A.D.brough t about large political, econo­ mic and religious changes in the region, and inaugurated the beginning of the end of the runoff farming hey-days in the Negev. The eastern part of the runoff farming district, where the churches ofAvda t were set on fire, may

37 have been abandoned more rapidly than the western area. It seems that runoff farming continued to function here for some time during theEarl y Arab period (Kraemer, 1958;Mayerson ,1960a ;Negev , 1981,Hairnan , 1982).

BEDOUIN FARMING , as semi-nomads, have sometimes engaged inrunof f farming, or very marginal farming with direct rainfall only,usin g the remaining fields of former periods,whic h they rarely repair.Ye t new terrace walls have also been built by Bedouin,especiall y in small wadis (Mayerson, 1960b). As this kind of farming is carried out by individual families and not planned by a central authority, it may well have occurred throughout certain historic periods, about which weca n only guess. In this context Mayerson (1960b:33)mad e an appropriate remark on runoff farming remains in the Negev ingeneral : "For the most part wear e dealing with humble field remainswhic h have a uniformity and anonymity resistant to identification".

MODERN ISRAELI RUNOFF FARMING Shortly before theStat e of Israel came intobein g in 19A8, the newly established ofRevivim , in the Yeroham -Revivi m zone, attempted to use runoff techniques inagriculture . These efforts inrunof f farming were also taken up by the settlers ofkibbut z Mashabbe Sade and kibbutz Sede Boqer in the early fifties. However, a pipeline with imported water from the north made pipeline-irrigation possible, which completely changed the course of agricultural development in these settlements (Hillel,1982) . Since 1958 three experimental runoff farms, near Shivta, Avdat and wadiMashash , wereesta ­ blished in theNege v by Evenari,Shanan ,an d Tadmor (1971, 1982).

HYDROLOGY OFRUNOF F FARMING

The major source ofwate r in the Negev hills isth e surface runoff from winter . The gentle nature of the rainfall regime in the area has already been described. Most rainfalls are in the range of0- 6 mm per day. Even exceptional rain-storms with more than 20m m ofprecipitatio n are usually composed of several individual showers, separated by sunny or windy periods (Yair et al., 1978).Rainfal l intensities equal to or greater than 20mm/hou r have an average cumulative duration of only 15minute s per year (Kutiel, 1978; Yair, 1983). Wide variations occur also in the areal distribution of

38 rainfall on smallwatersheds , due toth e influence oftopographi c differrences (Sharon, 1972). Hillside slopes receive twice the amount of rainfall recorded on peaks of hills and knolls in the Avdat area. These ridge and valley diffe­ rences are probably due to raindrops being blown away from peaks (Shanan, 1975). As soon as rain starts falling, it willhi t both the vegetation of a certain area as well asth e ground. Since vegetation isspars e in the region, interception storage issmal l and amounts to nomor e than 0.2 mm for a 12 mm rainfall (Evenari et al., 1975). What will happen to the rain that hits the ground, depends essentially on two factors: 1. The quantity ofrai n falling with time, i.e. the rainfall rate; 2. The amount of rainwater that can be absorbed by the ground with time, i.e. the infiltration rate. The latter factor isobviousl y dependent upon the nature of theground , which may be solid bedrock, bare soil,o r amixtur e ofbedroc k outcrops,stone s and soil. The initial infiltration rate ofploughe d loess soils inth e Avdat area, consisting of some 25% fine sand, 55 %sil t and 20% clay, is some 18 mm/ hour. This initial rate decreases fast and sharply to a final constant infil­ tration rate of about 2-3 mm/hour (Evenari et al., 1975), because a wet surface crust is formed. The puddling effect ofth e raindrops causes the wet upper loess layer to slake,a sver y fineparticle s and swelling clays clog the surface,thereb y greatly reducing infiltration (Hillel,1959 , 1964). The hillslopes inth eAvda t area are covered with amixtur e of loess and stones. These colluvial soils or lithosols have apermanen t loessial crust with an avearge thickness of 4.6 mm on a 1% slope, 3.8 mm on a 10% slope, and 2.6 mm on a2 0% slope. The initial infiltration rate ofthes e hillslope lithosols ismuc h lower than the 18mm/hou r measured incultivate d loess soils in the valley. Although of rare occurrence, it isnoteworth y that rainfall with an intensity ofmor e than 15mm/hou r breaks the soil crust, thereby causing an increase in infiltration (Shanan, 1975). Seasonal temperature also effects infiltration rates, ashighe r temperatures in autumn and spring cause a 1.8 and 1.4 fold increase, respectively. The threshold amount of rainfall needed to generate slope runoff in arid regions is rather low. Varying with conditions and rain proper­ ties, it amounts to 1-2 mm for massive bedrock and 3-5 mm for a stony soil cover (Yair and Klein, 1973; Shanan, 1975; Yair andLavee , 1976). In the latter case,mos t of thiswate r is used to wet and saturate the loessial crust and to seal its cracks (Shanan, 1975).

39 Photo 2. Glistening runoff generated on hillslopes in theAvda t area during a rain-storm, causing stream-flow inNaha l Zin below.

As soon as the rainfall rate is larger than the infiltration rate, the excess water first accumulates in little depressions,whic h are due to irregu­ larities of the ground surface. This initial loss of runoff water is called storage, and amounts to about 1m m in theAvda t area (Evenari et al., 1975). Ifth e rainfall continues at a higher intensity than the infiltra­ tion rate, a thin, shining sheet ofwate r is formed on the ground surface (Photo2) . This accumulated rainwater will now begin to run off on a sloping surface with a certain velocity as overland flow. Incas e of sufficient rainfall duration and intensity, the overland flow of runoff water will reach the first natural drainage channels. The velocity ofchanne l flow is larger than of overland flow,a n important difference to keep inmind ! Ifth e rainfall conti­ nues, the first order wadis are filled with runoff water and a real desert flood may develop, by which the runoff waters also reach the large wadis (Photo2) . In case of intermittant rain showers, water will evaporate from moist soil

40 surfaces. With resumed rainfall the soil surface needst ob e saturated again before runoff will appear. Evaporation is thus ahydrologica l loss from the system in runoff production. Measured with aUSW BClas s APan , evaporation on the Avdat farm-stead hill varied from 2.5 to 3.6 mm/day in the winter season (Evenari et al., 1971). Evaporation during day timewa s about 2 to 3 times higher than at night.A s described by Shanan (1975), evaporation from moist to saturated soils is approximately 0.7 times that of an open water surface ina USWB Class APan .Henc e evaporation from a saturated soil surface crust in the Avdat area in winter isbetwee n 1.8 to 2.5 mm/day. Shanan also studied the hydrological effect of the removal of stone cover from hillslopes upon runoff production. The ancient farmers in the runoff farming district used tochang e the natural stone cover onman y hillslopes by concentrating them intomound s and gravel strips (Evenari et al.,1971 , 1982). Shanan reached the conclusion that the removal of stones from the soil surface results in the forming of acontinuou s loessial crust, which is advantageous for runoff production inth e gentle rainfall regime of the Negev. The rainfall infiltration is initially larger on loess soils with a natural stone cover, because air isabl e to escape through vesicular layers beneath the stones. As the soil surface becomes more and more saturated with rain­ water, the air escape routes around theperipher y of the stones are sealed. When this threshold is reached, the overall effect ofth e stone cover on rainfall infiltration isreversed . Infiltration on a stone covered slope is now less" than on a loess slope cleared ofstones , because the stones are impermeable whilst the loess isn't. Since most rains inth eNege v are gentle and of rather short duration, this threshold isusuall y not attained, however, according toShanan . It has been measured by Shanan that stone clearance brings about an increase in average annual runoff production of 24% on 10% slope s and of nearly 50 %o n steeper 20% slopes (Shanan et al., 1969;Shanan , 1975). Apart from decreased infiltration under gentle showers, another positive effect of stone clearance might be considered. The reduced resistance to runoff overland flow perhaps leads toa n overall increase in its velocity. Thismigh t result in lower transmission losses and higher runoff yields, which could help to explain the dramatic increase in runoff yield from the steeper 20 %slopes . Average seasonal runoff yields in the area of theAvda t experimental runoff farm were investigated by Shanan during the years 1964/65 - 1968/69. Data from very small plots (natural stone cover), the southern man-made conduit catchments,an d the large natural eastern catchment arepresente d inTabl e 6.

41 TABLE 6. Average annual runoff yieldsi nth e catchment areao fth e Avdat experimental runoff farm, duringth e period 1964-65t o1968-6 9 (average annual rainfall during those years:94. 7m m *; basedo ndat a from Shanan, 1975).

CATCHMENT SIZE CHARACTERISTICS AVERAGE RUNOFF YIELD (naturalston ecover ) Peruni t Percentage * Absolute ofare a of rainfall quantity (mm ) (%) (cubicm )

OVERLANDFLO W

80m square 1% Slop e 26.3 mm 27.8 % 2.1 m3 80m square 10% Slop e 21.8 mm 23.0 % 1.7 m3 80m square 20% Slop e 11.4 mm 12.0 % 0.9 m3

OVERLANDAN DCONDUI TFLO W

1.0 ha Conduitcatchmen t 5 10.6 mm 11.2 % 106.0 m3 3.1 ha Conduitcatchmen t 4 5.4 mm 5.7 % 167.4 m3 3.7 ha Conduitcatchmen t 6 5.7 mm 6.0 % 210.9 m3 4.3 ha Conduitcatchmen t 7 5.8 mm 6.1 % 249.4 m3 5.4 ha Conduitcatchmen t 2 11.7 mm 12.4 % 631.8 m3 5.5 ha Conduitcatchmen t 3 4.3 mm 4.5 % 236.5 m3 7.0 ha Conduitcatchmen t 1 8.1 mm 8.6 % 567.0 m3

OVERLANDAN DWAD IFLO W

345 ha NaturalEaster ncatchmen t 0.5 mm 0.5 % 1725.0 m3

Small catchments generally yield more runoff per unito fare a than large catchments. The reaction time betweenth e onseto frainfal lan dth ebeginnin g of runoff production isals o shorteri nth e caseo f small catchments, as comparedt olarg e ones. Thesizabl e eastern catchmento fth eAvda t farm (345 ha) will only yield runoff unto the farm after relatively prolonged rainfall, whereas the smaller conduit catchments (1-7ha )wil l already produce some runoff after more gentle rain. However, when runoff does flowi nth ewad io f the eastern catchment, itusuall y contributesth elarges t amountso f water untoth efar mi nabsolut e quantities (Photo 3). These variationsi n reaction time, frequency offlow ,an dabsolut e quantity ofrunof f produced, mustal lb e considered inprope r runoff farming designan dmanagement . The conduits that divideth esouther n catchmento fth eAvda t farm into smaller sub-catchments (Evenarie tal. , 1971,1982 )wer e builtb y mani n

42 !!s-: '^ 1

: '•"•''";, 'i '""'[ iff ,

-«- "^"'^. :,, "^•B'fc^Wil ;'

PHOTO 3.Larg e quantities of runoff water flow through thewad i of the sizable eastern catchment (3A5ha )int o the Avdat farm. antiquity (the hillside conduit system, p. A9), presumably for some of the hydrological reasons stated above and below. The differences in runoff yields between the seven sub-catchments of the southern drainage basin are not only determined by individual catchment size, as isclea r from Table 6. Variations in surface cover, vegetation,slop e angle,lengt h ofoverlan d flow, landscape irregularities, quality of the conduits,etc .hav e their respective influence upon the ultimate runoff yield. Runoff from catchments is, according to Shanan (1975), adversely affected by overland flow length.A n increase in the length ofoverlan d flow from 20 to 100 m causes a5 0% reduction in runoff. Small catchments are likewise more efficient producers of runoff. Shanan considers initial water losses, i.e. rainfall required to initiate and transmit runoff, asth e most important reasons for these differences. Initial water losses in the investigated catch­ ments amount to 2.5 mm for the small plots of8 0 square meters,5. 5 mm for the southern conduit catchments of 1t o 7hectare , and 7t o8 m m for the third order eastern drainage basin of 3A5 hectare. The initial losses in the latter

43 basin are accounted for by Shanan in the following manner: crust wetting 2.5 mm,overlan d flow losses3. 0 mm,wad i channel losses2. 5 mm. These figures from the Avdat area in the Negev are comparable to those found for parts ofNort h Africa. A minimum rainfall of5 m m is required to initiate runoff inth eMza b basin inAlgeria , and inEgyp t a rainfall of 8m m is necessary to cause a flood (Dubief, 1953;qf .Shanan , 1975). The flow velocity ofshallo w runoff overland flows is a key to grasp the significance of the concept of partial area contribution (Yair et al., 1978). In a rough stony environment like the Negev hills runoff flow velocities are low, being in the order of3- 6 cm/sec, seldom exceeding 10cm/se c (Emmett, 1970; Lavee, 1973). Asrainfal l duration isusuall y short in the central Negev hills, surface runoff generated at the upper part ofslope sexceedin g some 50 m in length probably has little or no chance of reaching the slope base. Only the lower part ofth e slopes may contribute runoff to a related wadi channel during most rainfall events (Yair et al, 1978). This principle was apparently well understood by the ancient runoff farmers who designed the conduit system to overcome this particular problem in certain landscape situations. A large catchment can be divided into several smaller ones through the building ofconduits , thereby shortening the length ofover ­ land flow considerably. Since channel flow in acondui t ismuc h faster than overland flow, more runoff water can thus be harvested from the hillslopes unto the arable fields (p.49 ,phot o6) . Whereas the runoff farmer aims to minimize infiltration in the runoff producing-parts ofhi s catchment(s), his strategy changes completely when the runoff water finally flows into the farm lands proper.Her e the precious water has to be conserved asgoo d aspossible . The fields are therefore terraced, while check-dams retain the harvested runoff water. Notwithstanding the positive effect ofploughing , infiltration of the runoff water into the loessial soil is a slow process and usually takes one to three days. The height of the lowest part ofth e check-dam above the ground level,ofte n made up by a spillway, determines the amount ofwate r that is kept on the field. According to experiments, 1m m ofwate r suffices to moisten 8-10 mm of loessial soil to itswater-holdin g capacity. If 300m m of water isretaine d by acheck-dam , a loess soil of2 to3 mete r deep will be fully moistened. After thewate r has completely infiltrated into the soil,th e surface crust will dry out in about 24-36 hours. Once the crust isdry , the soil is rather well sealed to keep themoistur e inside for agricultural pur­ poses (Evenari et al., 1971).

44 RUNOFF FARMING SYSTEMS

A proper understanding and analysis ofth e landscape is an essential prerequisite in the design of runoff farming systems. Apart from theclimate , the landscape must also be suited for rainwater-harvesting agriculture. Man sometimes needs to reshape or alter the landscape inorde r to improve the setting for runoff farming. The following minimal requirements have to be present inth e landscape: a) Geomorphic surface conditions must be such that runoff isrelativel y easy generated byrainfall . b) Differences in elevation must be present, so that runoff waters can be concentrated in the lower parts of the landscape. c) The runoff receiving landscape parts must have sufficiently deep soils,o f suitable texture and structure, to retain and store the received runoff water for agricultural purposes.

Runoff farming systems were classified by the author (Bruins et al., 1986a) in relation to specific landscape properties, aseac h runoff farming system has tob e adapted toa certai n geomorphic situation:

1. Micro-catchment system 2. Terraced wadi system 3. Hillside conduit system 4. Liman system 5. Diversion system

The order in which the five systems distinguished here are described isno t arbitrarily chosen. Itbegin swit h the system of the smallest geomorphic scale and continues successively with increasingly larger systems in landscape settings of greater dimension (Bruins et al.,1986a) .Fro m ahistori c perspec­ tive, it should be noted that the micro-catchment system isapparentl y a new development in the Negev, having been introduced inth e last 25year s of this century (Shanan et al., 1970; Evenari et al., 1971, 1982). A kind of liman system already functioned in the area in antiquity, though detailed case studies are lacking.Thi s latter system isuse d today inth e Negev inparticu ­ lar to establish green areas along the roads, planted with trees that provide both shade and improve the architecture of the landscape.

45 1.MICRO-CATCHMEN T SYSTEM.

Rather flat or slightly undulating often lack a natural distribution of runoff-producing and runoff-receiving areas. These type of landscapes seem, therefore,les s suited for rainwater-harvesting agriculture. However, ifth e soil has the right texture (gravel or course sand areou t of the question) and iscapabl e of producing runoff, then micro-catchments (Shanan et al., 1970;Shana n and Tadmor, 1979)ma y be introduced successfully. Micro-catchments have to be made artificially by man, involving earth movements to create arunoff-producin g area and arelate d infiltration basin. The size ofmicro-catchment s is generally lesstha n one hectare, whereby the infiltation basin, usually alarg e pit (Photo 4), isdu g at the lowest spot.

PHOTO A. Micro-catchment with fruit tree inth e Avdat area. The infiltration basin is filled with runoffwate r from the glistening runoff-producingpart .

46 In aver y flat landscape relatively more work isrequire d tomak e arunoff - producing and runoff-receiving area,i n which trees,pastur e or food crops can be planted. When even slight natural slopes exist,an d the surface is capable of producing runoff, then a runoff-producing area isalread y present. It is important to design and built the micro-catchments according to the best fit with the existing landscapecontours . Crop requirements also influence micro-catchment design. Square,rectangu ­ lar or circular micro-catchments are generally better suited for trees, although food crops orpastur e plants can also be grown inthi s way, appearing as little green islands in the landscape. Longitudinal micro-catchments, following the contour lines, enable mechanization and areofte n preferred for field crops. Various types and shapes ofmicro-catchment s are reported in the literature, reviewed by Boers and Ben-Asher (1982), who define micro-catch­ ments as having a runoff flow trajectory of lesstha n 100 m from runoff contributing area toth e receiving area or infiltration basin. Of paramount importance isth e relationship in size and design between runoff-producing and runoff-receiving area. Enough runoffwate r must begene ­ rated and stored forprope r agricultural production. The above relationship depends upon climate, geomorphology, soil and type ofcrop . A noteworthy advantage of micro-catchment systems is the high specific runoff yield as compared with larger catchments (Shanan, 1975; Shanan and Tadmor, 1979). A disadvantage isth e low number of crops per unit area and the enhanced asa result .

2. TERRACED WADI SYSTEM.

A terraced wadi system involves the building of aserie s of low check- dams across a wadit o retain the runoffwater s in the stream channel and enable the water tob e stored in thewad i soil. The check-dams or dikes were built from local-stone spu t together without cement in single-line ordouble - line walls.Eac h individual terrace behind acheck-da m ought to be levelled to ensure proper water spreading and storage inth e soil. This terracing of the wadi had the" effect oftransformin g acertai n part ofth e valley into a continuous stairway, with stairs perhaps 10-20meter s wide and 20-50 cm high (Hillel, 1982). Hundreds of terraced wadis can be found in the runoff farming district. Suitability ofa wad i for this system depends mainly upon the follo­ wing three factors:

47 a) Hydrology ofth ewad ian d itscatchment . b) The ratio between catchment area and thewidt h (area)o f the wadi bed, in which the agricultural fields are to be located. c) Soil depth and texture ofth e wadi bed.

The terraced wadi isprobabl y the most ancient type of aman-mad e runoff farming system inth e central Negev hills by which stream valleys were changed considerably. The profile of a terraced wadi differs markedly from the origi­ nal shape of the stream channel. Erosion and deposition processes were pro­ foundly changed as aresul t of terracing. A number of terraced wadis seem to date back to the Israelite Period, nearly 3000 years ago (Evenari et al., 1958; Aharoni et al., 1960; Cohen, 1976). The system does not appear to be confined, however, to a specific age. The investigated wadi ofNaha lMitna n seems to have been terraced in Byzantine times. Bedouin have also engaged in the terracing ofsmal l wadis,accordin g toMayerso n (1960b).

PHOTO 5.A terraced wadi, now being eroded, upstream from the Shivta farm.

48 Small and medium-sized wadis, situated in rather narrow valleys,ar egene ­ rally most suited forthi s system of runoff farming. The runoff-producing hillslopes ought to rise dirctly from the sides of the terraced wadi bed. The possible presence ofcolluviu m is aclea r disadvantage, which may prevent runoff from reaching thewadi . The terrace dams in the smaller wadis usually lack aspillwa y and the floodwaters were allowed tooverto p the entire length of the check-dam. Inth e somewhat wider wadis,however ,a specia l spillway was built in the check-dam topreven t erosion. The spillway overflow level from one wadi terrace to the next may be 15t o 30c m high, depending upon the average frequency and magnitude of runoff events aswel l asupo n the crops to be grown. A spillway level of2 5c m above the field ensures theretentio n and storage of about 250m m ofwate r when runoff has filled thewad iterraces .

3.HILLSID E CONDUIT SYSTEM.

The hillside conduit system is ausefu l rainwater-harvesting agricultural system in geomorphic situations,b y which rather wide stream valleys with good agricultural soils are surrounded by hills having long slopes or colluvium (Bruins et al., 1986a). This system is found regularly in the central Negev hills and seems todat e back to the Nabatean-Byzantine period. When the length ofrunof f overland flow over a slopei s long, water is being lost (Shanan, 1975)o r may not even reach thevalle y fields at all due to the low velocity ofoverlan d flow (Yair et al., 1978). Channel flow,o n the other hand israthe r fast and very little water islos t en route. By building simple ditches asconduit s on the hillslopes, the length of overland flow isconsiderabl y shortened. Thus anumbe r of sub-catchments are formed. The overland flow ineac h sub-catchment isdetermine d by the spacing of the conduit channels. The conduits themselves are low walls, simply built from loose stones and soil, having aslop e of nomor e than 1% . The conduits also enable the runoff tobypas s possible colluvial deposits,whic h otherwise might have absorbed (Yair, 1983)par t of the runoffwaters . In anumbe r ofgeomorphi c situations more runoffca n beharveste d through the introduction of conduits. The runoff waters drained by the conduits from the respective sub-catchment (Photo 6)ca n be directed specifically towards certain agricultural fields, or into cisterns and other storage reservoirs. This regulating aspect of conduit channel flow isanothe r advantage of the system.

49 PHOTO 6. Runoffwate r drained by a hillside conduit flows into the Avdat farm.

The valley soils that receive the runoffwate r from the conduits for agri­ cultural purposes must obviously be terraced, whereby check-dams and spillways are used to regulate, retain and store the harvested water.Th e Avdat experi­ mental runoff farm (Evenari et al., 1971,1982 )exhibit s a nice example of the hillside conduit system (Photo 6), which ismor e complicated and generally of a larger geomorphic scale than the terraced wadis.

4.LIMA N SYSTEM.

Wherever tributary wadis, known to carry runoff floodwaters in due season, widen or come unto a large plain with good soils, limans may be constructed. A check-dam isbuil t around the levelled to retain the runoff waters from the wadi. A spillway regulates the level ofth e water in the liman to prevent the destruction of the check-dam. In larger wadis a series of limans may bebuil t after each other. When the first liman has filled up, thewate r will flow through the spillway into the second liman and so forth,provide d enough runoff water isgenerate d inth e catchment area.

50 **<'S i *i»" % -i jm ^

.-tff= ! _ .itjgf

PHOTO 7.Liman s filled with runoff water at theWad iMashas h farm.

The term liman isderive d from the Greek word "LIMNE" which means . A liman just filled upwit h runoff water looks like a lake indeed (Photo 7). The catchment size of alima n isgenerall y larger than in theprecedin g systems, though not always.Mini-liman sca n also bemad e (Bruins et al., 1986a). At theWad iMashas h experimental runoff farm in theNegev , altogether 10 hectares of limans have been established, planted with fruit and fuel trees (Evenari et al., 1982). Most ofth e limans function ina successfu l manner. Many more limans have been constructed in theNege v by the authorities, gene­ rally along the main roads for landscaping purposes.

5.DIVERSIO N SYSTEM.

In landscape sitiuations where large areas ofpotentia l arable land are found at anelevate d position with respect toth e stream bed of an adjacent wadi, it becomes necessary to build more intricate structures in order to divert the runoff floodwaters onto the fields. For thispurpos e ada m must be constructed in the wadi, upstream from the planned fields, at apoin t where

51 the level of thewad i bed Is at a sufficiently higher elevation than the fields under consideration. At this point a dam isbuil t to raise the flood- waters. A low threshold dam may often be sufficient. A diversion channel is excavated from the dam towards the planned agricultural fields. The gradient of the diversion channel must obviously be less than the actual wadi and the adjacent terrace plains.I fthi s precondition is successfully implemented, the diversion channel will emerge further downstream untoth e surface of the arable lands,which , from hereon,ca n be irrigated with diverted floodwaters. Three ancient diversion systems in the runoff farming district,alon g Nahal Lavan,Naha l Avdat,an d Nahal Mamshit,wer e studied in detail by Shanan et al. (1961). The largest and most striking example of adiversio n system in the Negev is that ofNaha lLavan , near theNabatean-Byzantin e town ofShivta .Th e terraced agricultural fields comprise here an area of about 200 hectares, showing massive ancient walls with spillways of 30-60mete r in length.

PHOTO 8.Moder n diversion channel,70 0m long, departing from Nahal Avdat onto a cultivated area of6 ha, amidst many remains ofancien t runoff farming, as well asmoder n micro-catchments,belongin g to theAvda t runoff farm.

52 PHOTO 9. Runoff water flowing from this diversion channel into the farm lands.

The handling of large flows of diverted water in these diversion systems required undoubtedly greater skill and sophistication than the operation of small runoff farms. Maintenance ofthes e systems often became increasingly difficult due to silting up problems, necessitating the construction of new diversion structures,beginnin g higher upstream, to ensure the gravity flow of diverted floodwaters toward the fields (Shanan et al., 1961;Hillel , 1982). In the last ten years adiversio n system has been set up near the Avdat experimental farm, constructed by Mazig (Evenari et al., 1982). An approximately 50 cm high threshold dam was built at an appropriate place in Nahal Avdat. From this dam a 700mete r long diversion channel was excavated into theLat e Pleistocene loessial terrace deposits ofNaha l Avdat (Photo 8). Since the diversion channel has a smaller gradient than the wadi and its adjacent terrace plain, the channel bottom reaches the surface ofth eculti ­ vable lands after about 700m (Photo 9).A loessial serozem soil, exhibiting a well developed calcic horizon,ca n be seen in certain parts along the sides of the diversion channel. When there isa flood in Wadi Avdat, the threshold dam raises the level of

53 the water, enabling the precious liquid to flow into the diversion channel. The diverted floodwaters flow through the diversion channel and finally emerge onto the farm lands ofth eLat e Pleistocene loessial terrace of Wadi Avdat (Photo 9). In addition, runoff water from a tributary wadi ofNaha lAvda t is also conducted onto the same arable lands. The agricultural fields are sur­ rounded by check-damsan d interconnected via spillways. An area of6 hectare of desert land iscultivate d in thismanner ,plante d with pistachio and almond trees.A desert park has also been laid out for recreational purposes (Evenari et al., 1982). Micro-catchments surrounded by low earth bundswer e constructed alongside ofth e diversion channel, particularly in the terrain that gradually descends towards the channel. The infiltration basins ofth e various micro-catchments appear as little dots onPhot o 8. Apart from the agricultural objectives,th e micro-catchments help to check erosion caused by undesired runoff flows from this sloping land toward the side of the channel. Inthi smanner , different runoff farming systems areno t only adapted to local conditions, as dictated by the geomorphicmake-u p of the area, but also may protect each other in erosion control.

54 4 Pastoralism

NOMADIC PASTORALISM AS A SPECIALIZED FORM OF A FOOD-PRODUCING ECONOMY

Pastoralism hasbee na nimportan t formo fland-us ei nth e arid Negev during historical times. However,th e practisingo flivestoc k herding,eithe r bya nomadic , semi-nomadico rsedentar y population,i softe n hardt ogras pi n archaeologic and historic terms. Asa food-producin g system in the area, pastoralism cannotb edisregarde d inth e contexto fthi sdissertation .Beside s the available data from theNegev , mentioned below, the originsan dvariou s forms ofpastoralis mi nth e regiona tlarg e are treatedi na somewhat wider historican dgeographi c perspective. Nomadican dsemi-nomadi c livestock rearing hasbee n practised widelyi nth e great arid belt that stretches from theAtlanti c coasto fth eSahar a toth e ofMongolia . Interm so fman-lan d relationships extensive livestock rearing often developedi nmargina l environments where food production through crop cultivationi seithe r impossibleo ra rathe r risky undertaking.Henc eth e natural vegetationi sharveste d throughth egrazin g animalsan di sthu strans ­ formed intomilk , meatan dothe r useful products forman . As such, mobile extensive pastoralism canb efull y regardeda sa specialize d formo fa food- producing economy (Khazanov, 1985). There are, however, alarg e numbero fpossibl e interrelationships between livestock rearing and crop cultivation.Thi srati oo fpastoralis man d agricul­ ture not only depends upon the environment, butals o upon various economic, social andpolitica l factors. Theinfluenc eo fthes e latter aspectsi sexem ­ plified byth efac t that pastoral nomadsma yliv ei narea sfi t for cultiva-

55 tion, while not engaging significantly inagricultur e if at all (Cunnison, 1966; Marx, 1982). "Pure" pastoralists can only develop within a complex economic system, ofwhic h they are an integral part.The y produce luxury food for an urban or well-to-do rural population.Thi s kind ofpastoralis m is quite distinct from "subsistence pastoralism", which isusuall y a combination of livestock rearing and farming (Marx, 1986,persona lcommunication) . Pastoralists hardly ever subsist directly on their animals,a s far as their diet isconcerned . Ifthei r food was only composed ofeatin g their livestock and its products, they would require very large herds indeed (Marx. 1982). Pastoral specialization hasmean t more or less economic one-sidedness and no autarky, as pointed out by Khazanov (1985:3)i n anexcellen t and comprehensive study about nomads and the outside world. He states that nomads could never exist on their ownwithou t the outside world and its sedentary societies and different economies. Marx (1982)argue s that animals areno t merely produced in apastoralis t society,becaus e they contribute toth e diet of their owners, but rather for saleo r for exchange against grain or other goods. Cereals often constitute the staple food ofman y pastoralists. Khazanov (1985:16)define d the most important characteristics of pastoral nomadism:

1.Pastoralis m isth epredominan t form of economic activity. 2. Itsextensiv e character isconnecte d with themaintenanc e ofherd s all year round on a system of free-range grazing without stables. 3.Periodi c mobility in accordance with the demands ofa pastora l economy in specific grazing territories. U.Th e participation in pastoral mobility ofal l orth e majority of the population. 5.Th e orientation of production towards the requirements of subsistance. (Khazanov remarks that this fifth characteristic today no longer applies in many cases, as pastoral nomads have been drawn into the world market system.)

The most important problem ofpastora l nomadism isth e balance between availability ofnatura l resources, number of livestock and population-size (Barth, 1964;Khazanov , 1985). The thesis ispu t forward by Barth (1959/60:8) that: "Unless techniques for the storage of fodder are developed, absolute population-size is limited by the carrying capacity ofth e pastures in the least productive period ofth e year." (cf. Khazanov 1985:70). Not only the

56 intra-annual shortages ofpastur e should be taken into account, but also the Inter-annual frequency of drought years. Pastoralism Inth ecentra l Negev without any feed reserves or supplementary supplies has led to , in the absence ofa coordinate d policy. In good years the flockswoul d multiply, and in drought yearsthei r numbers would exceed the dry range'scarryin g capacity. During a succession ofdrought s the Bedouin flockswoul d bedecimated , while their ownerswoul d be impoverished for years. Their plight could be so desperate that the semi-nomads or nomadic desert dwellers, driven by starvation,woul d enter morehumi d sedentary areas in search for food (Hillel, 1982). An example of such an incident, although not necessarily related to a drought, is reported byBaile y (1980) to have occurred during the Ottoman period in 1877:"troop s fired upon the bedouins as they were carrying offcrop s nearGaza... " In a larger perspective, that includes Africa,L e Houerou (1985)i spessi ­ mistic about the present impact ofma n in pastoral areas. Many factors, such as overstocking, wood cutting,huma n and animal population expansion,contri ­ bute to intensify thepressur e on grazed ecosystems. These, in turn, become less resilient and lessabl e to absorb climatic droughts, which they used to do under the light exploitation conditions of thepas tcenturies . Le Houerou (1985), who wrote the article in 1982, remarked that if this situation persists, one can only predict agloom y future for pastoralism. will either have to disappear quietly or evolve towards other systems ofanima l production with increased reliance on supplementary feedings for longer and longer periods. Khazanov (1985:71,76) is equally pessimistic: "Pastoral nomadism isdoome d to stagnation because itseconom y is extensive... Increased production innomadi c society asa whol e is, to any significant extent,impossible" .

PASTORALISM AND AGRICULTURE

Pastoral nomadism proper is in itsmos t puremanifestatio n characte­ rized by the total absence of agriculture. Semi-nomadic pastoralism is much more widespread throughout the world. Itspredominan t activity isals o com­ posed of extensive pastoralism and the periodic changing of pastures, but agriculture ispractise d aswel l in a secondary and supplementary capacity. There are, however, many other forms ofpastoralism ,whic h are somehow inte­ grated with agricultural practices.Althoug h there seems to be aclea r lack of

57 detailed historic and archaeologic information about these relationships in the Negev inantiquity , some of the systems described by Khazanov (1985:19) may actually have existed here in the past, used either by the Bedouin (Marx, 1967)o rb y the sedentary runoff farmers:

1) SEMI-SEDENTARYPASTORALIS M differs most fundamentally from semi-nomadic pastoralism in the sense that agriculture plays thepredominan t role in the general economic balance.

2) HERDSMAN HUSBANDRY orDISTAN T PASTURE HUSBANDRY describes the situation in which the majority ofth e population leads a sedentary life and is occupied for themos t part with agriculture. The livestock or,mor e often,som e ofit , is tended all year round by herdsman especially assigned to this task. They maintain the herds onpastures , sometimes quite far from the settlements.For part ofth e year the animals are usually kept in enclosures, pens and stalls, which entails the laying-in of fodder.

3) SEDENTARY ANIMAL HUSBANDRY usually supplements agriculture and has diffe­ rent variants. a) HOUSEHOLD-STABLE ANIMAL HUSBANDRY ison e of these variants. It is characterized by thegrazin g of livestock in pastures adjacent to the settle­ ment forpar t ofth eyear , whereby the animals usually return every day. For the other part ofth eyea r the livestock are kept in stables and enclosures and fed accordingly. b) SEDENTARY HOUSEHOLD HUSBANDRY WITH FREE GRAZING isye t another variant of sedentary animal husbandry. This is genetically one ofth emos t primitive and ancient forms ofpastoralis m inwhic h the laying-in of fodder and the maintenance of livestock inenclosure s or stables isgenerall y absent or very limited (Khazanov, 1985).

The technical problems and organizational requirements of a possible inte­ gration between runoff farming and animal husbandry in thecentra l Negev in antiquity are eloquently described by Hillel (1982:187): "All the evidence at hand leads to the conclusion that the ancients, espe­ cially during the Nabatean-Roman-Byzantine epoch, maintained a diverse agri­ cultural economy inth e central Negev, with cultivation and grazing carried out simultaneously. Since roaming livestock could devour and trample crops, destroy runoff-collecting conduits, accelerate erosion of slopes, and induce

58 silting and pollution of cisterns, undisciplined grazing has always been a menace to farming.Onl y an organized and coordinated society could resolve and integrate such potentially contradictary activities as farming and grazing,a s well as land and water rights,s oa s to avoid conflict and damage." No such coordination existed, in the opinion ofHille l (1982), from the Early Arab period until the establishment of the State of Israel. The tribes of Bedouin that inhabited the region could not be certain of long-term land ownership, because ofintermitten t struggles over territorial rights. Bailey (1980) collected material about such situations of internal warfare between the Bedouin tribes during the 19th century, based upon oraltradition s ofth e Bedouins themselves.

THE ORIGINS OFPASTORALIS M

Most authorities agree that pastoral nomadism did not evolve from hunting and food gathering societies. It rather developed asa specialization out of sedentary agriculture (Grigg, 1974). Itssource s go back to the Neolithic revolution and the emergence of a food-producing economy, which basically always consisted of two forms of economic activity:cro p cultivation and animal husbandry.A spointe d out by Khazanov (1985), only groups leading a relatively sedentary way of life with definite surpluses of vegetable food at their disposal could domesticate animals. In a study about the emergence of specialized pastoralism in the southern , Levy (1983)point s out that during theLat e Neolithic, some 6000-7000 years ago, crop cultivation and livestock grazing in thenorther n Negev is clustered around springs, in year-round grazing areas. Themaximu m grazing range of away from these springs is about 5-8 km. Thismigh t be consi­ dered as a form ofhousehol d animal husbandry with free-range grazing, as described above. With population growth and the development of ane w agro-technology, the subsequent Chalcolithic settlements moved further intoth emor e arid inland valleys, probably using runoff floodwaters for crop cultivation. Itwoul d have been aphysiologica l necessity for village communities inth e inland valleys tomov e sheep or herds to the more humid coastal plain on the arrival of the dry summer season. The temporary nature ofChalcolithi c settlement on the coastal plain indicates, according toLevy , the primarily pastoral nature of these camps aspar t ofa newl y developed subsistence strategy, inwhic h sheep

59 and goat herds owned by permanent villages were moved north-westward for grazing during the late summer and early winter months. This development of sedentary pastoralism during theChalcolithi c (4000-3200B.C. ) marked the emergence of the traditional system of land-use and herd-management in the Northern Negev,whic h matured during theEarl y Bronze Age (3200-2200 B.C.) and lasted until the 20th century (Levy, 1983). This system has the following characteristics:

1. Grazing by mobile herds of small livestock (sheep and goat)combine d with the dry-land cultivation ofwinte r cereals (wheat and barley)wheneve r and wherever possible. 2. No significant use of irrigation water, fertilization or supplementary feeds. 3. The grain and livestock products are primarily used for the subsistance of the local population,althoug h surpluses are traded or exchanged in favou­ rable years (Noy-Meir and Seligman, 1979;cf .Levy , 1983).

The adaptation of a food-producing economy to a specific habitat led, especially in arid zones (LeHouerou , 1985)t o the predominance of pastoralism over other forms of agriculture. Apart from social and political factors, it can be said (Khazanov, 1985:69)tha t pastoral nomadism and semi-nomadic pasto­ ralism developed and functioned inth e first instance inthos e regions where they had economic advantages over all other kinds and forms of economic acti­ vity. Khazanov considers that climatic changes played an impor tant role in the emergence ofman y types ofpastora l nomadism, but economic and cultural preconditions were also necessary. The was themajo r centre ofnomadizatio n in south-west Asia and its role was similar to that of theSahar a to the development and dissemination ofAfrica n nomadism. Just as inth eSahara , appears in Arabia in the sixth millennium B.C. and small stock appears in the fourth and third millennia B.C., when pastoralism becomes the predominant economic acti­ vity. Sheep and goat definitively replaced cattle inCentra l Arabia from the end of the third millennium B.C. Herdsman husbandry and semi-nomadic pastora­ lism existed in theNea r East during the third and second millennia B.C., but in no way was there real nomadism. Theclimat e ofArabi a altered at about the same time asth e climate in the Sahara did (McClure, 1971). It isno t known how the desiccation which began about 2500 B.C. and grosso modo continues until thepresen t (Pearse, 1971;

60 Maley, 1973; Flohnan d Nicholson, 1980;Nicholso n andFlohn , 1980) affected the ancient pastoralists of Arabia. Perhaps one consequence of it was the movement ofpastoralist s to the borders ofagricultura l areas (Oates, 1976). What isclear , however, istha t the "bedouinization ofArabia " or the final stage of the formation of real pastoral nomadism inArabi awa s linked to the beginning of the utilization of the asa riding animal (Khazanov, 1985): "It was only specialized camel-herding, which began inArabi a somewhere in the middle of the second millennium B.C. and was possibly stimulated by the desiccation of the peninsula, which led to the dissemination of real pastoral nomadism inth e inner regions ofArabi a and Syria. Already in the eleventh century B.C. theMidianites , Amalekites and "children of theEast " stormed through Jordan intoPalestin e (Judges 6:1-6), "...for both they and their were without number" (Judges 6:5). According toDosta l (1959:22)thei r invasion was linked toth e perennial problems faced by nomads: the desire to extend grazing territory for the increasing numbers inthei r herds, the need of agricultural products and of trade with sedentary societies, etc." (Khazanov, 1985:100). Le Houerou (1985)wrot e an excellent article about the impact of climate on pastoralism, dealing with specific quantitative aspects ofpastora l man-land relationships. The development of pastoral nomadism inth e is summarized by Khazanov (1985:102) inth e following way. Nomadic pastoralism evolved out of amixe d economy ofcro p cultivation and animal husbandry, passing through the stage of extensive herdsman husbandry and/or semi-nomadic pastoralism. The earliest formso fextensiv e pastoralism were inth e first instance based upon the herding of small livestock. Thecame l gradually gained an important position in the economy as food, aswel l as for transport and riding purposes. The immediate transition topastora l nomadism in itstradi ­ tional economic forms, at least with regard to species-composition of herds and methods of their utilization, was probably linked to specific changes or variations of climate. Finally, nomads were linked right from the beginning in a complex system ofpeacefu l and hostile relations with sedentary states (Khazanov 1985:102). The rather scanty historic information about the interrelationships in the Negev between nomadic or semi-nomadic Bedouins and the sedentary runoff far­ ming population, including travellers, is described byMayerso n (1963, 1964) for theByzantin e andEarl y Arab period. Marx (1967, 1977, 1982)an d Baily (1980) have studied contemporary Bedouin life inth eNege v and its recent history.

61 5 Horvat Haluqim

ENVIRONMENT

GEOMORPH0L0GYAN DGEOLOG Y The siteo fHorva t Haluqim exhibitsa classi c exampleo fa nancien t terraced wadi system (p. 47),constructe d forrunof f farming purposes well before theNabatea n period,a sindicate db yarchaeologica l excavations (Cohen, 1976) and investigationsb yth eautho ri non eo fth e terraced fields.Th e remains of the ancientagricultura l villagear esituate d on the southern slopeso fth eHaluqi m anticlinal range, some4 k mt oth e northo fth ecanyon ­ likeZi n Valley which constitutes themos t dramatic geomorphic featurei nth e area. Thehill so fHa rHaluqim , risingt oa naltitud eo f61 8m , arepar to f theSouthwest-Northeas t trending fold beltso fth enorther n Central Negev. Twomoder n settlements, Kibbutz SedeBoqe ran dMidreshe t Ben-Gurion, are situated to thesout ho fHorva t Haluqimi nth eadjacen t synclinal plain of SedeZin , ata naltitud eo f470-47 5m .Thei r respective distancest oth e site are about1. 5k man d4. 5km . Theroa d from SedeBoqe rt oYeroha m formsth ed e facto southern bordero fth e site. Thesynclina l plaino fSed eZi ni slargel y covered by Late Quaternary loessial deposits (Issare t al., 1984)o fth e NetivotFormatio n (Bruinse tal. ,i npreparation) . The siteo fHorva tHaluqi m comprises three small tributary wadis (Photos1 , 10, 12; Fig.9 )o fNaha lHaroa ,whic h drain parto fa hil l composedo fUppe r Cretaceous carbonate rocks. Thewadi sar eo fth e firstt osecon d order and begin ata naltitud eo fabou t 530-540m a tth euppe r slopeso fa rathe r flat, east-west trending hillto paxis ,whic hi sabou t 200m wide .Th elengt ho fth e wadis from their inception untilth eroa di si nth erang eo f50 0t o 600 m.

62 Their average width ranges from less than5 m u pt o25-3 0m . Ini­ tially thestrea m bedsar e narrow asth ewadi s descend from about54 0 to52 0m . Whenth ewad i beds widen and their gradient diminishes, ancient agricultural terraces ap­ pear,usuall y beginning ata nalti ­ tudeo fsom e 515-520m . Henceforth the wadisar eterrace d until the vicinity ofth eroad , whilst their beds gradually descend toa nalti ­ tudeo fabou t49 5m . The lithology of the hill drained byth ethre e wadisi sdomi ­ nated by theNeze r Formation of

.500 m Late Turonian age,characterize db y TTJTTT Escarpment well bedded lithographic limestone Wadi * Spring and minor chalk layers (Arkin and FIG.8 .Locatio no fHorva t Haluqim. Braun, 1965), having asouthwar d dipo fabou t6 . Thenatura l join­ ting pattern andth eindividua l thickness (about3 0cm )o fth ehar d micritic limestone layers makesth eNeze r Formation suited forloca l building stones that areeasil y quarried andhandle d (Photo 11), andwhic h were used in the past. Turonian rockso fth eunderlyin g ShivtaFormatio nar eexpose d alongth e lower parto fth estream-valle y slopes, aboveth ethre ementione d wadis. The Shivta Formation consistso fpoorl y bedded calcarenitic limestone layers.

CLIMATE The site issituate d inclos e proximity toth enearb y meteorological sta­ tion ofSed e Boqer. Climatic data fromth elatte r station, belonging toth e Israel Meteorological Service,ar equote d from Arzi (1981).Th eaverag e annual rainfall is9 4mm , measured inth eperio d from 1951/1952t o1979/1980 .Inter - annual variations fluctuated froma minimu m of30. 9m mt oa maximu m of 167.3 mmpe ryear . Theaverag e number ofrain y days amountst o 26pe r year; the range of annual variation liesi nbetwee n 11an d3 6rain y days. The yearly temperature average is18. 2degree s Celsius.Averag e daily temperatures inth e coldest month (January)rang e from 3.9 Ca tnigh tt o15. 4 Ci nth e daytime,

63 lAbLL 7. CLIMATIC DATA FROM THESED E BOQER METEOROLOGICAL STATION.*

OCT NOV DEC JAN FEB MAR APR

RAINFALL (mm) 2.6 9.5 22.4 23.3 15.7 11.9 7.2 RAINYDAY S 0.8 2.8 5.3 6.5 4.3 3.8 2.0 TEMPERATURE (°C) 20.6 16.0 11.4 9.6 11.6 13.6 17.0 RELATIVEHUMIDIT Y (58) 60 62 68 70 64 58 53 EVAPORATION (mm/day) 6.0 4.0 2.9 3.1 3.8 5.1 7.3 EVAPORATION (mm/month) 186 120 90 96 107 158 219 WINDDIRECTIO N N N NE W W N NW

ANNUALTOTA L MAY JUN JUL AUG SEP ORAVERAG E

RAINFALL (mm) 0.7 0.2 0 0 0 93.5 RAINYDAY S 0.4 0.1 0 0 0 26 TEMPERATURE (°C) 20.9 24.2 24.8 25.3 23.6 18.2 RELATIVEHUMIDIT Y (%) 43 45 53 57 62 58 EVAPORATION (mm/day) 9.4 11.0 10.7 10.1 8.3 6.8 EVAPORATION (mm/month) 291 330 332 313 249 2491 WINDDIRECTIO N N NW NW NW NW NW

*Modifie d from Arzi (1981).

All dataar eaverag e figures,derive d fromth eIsrae l Meteorological Service (Climate Department): Rainfall data andth enumbe r ofrain y daysar efro ma 2 9 year period (1951/1952t o1979/1980) . Temperature andrelativ e humidity data are from a1 0yea r period (1952-1956 and1964-1969) . Evaporation datao f an open water surface were measured ina Clas sA Evaporation Pan, up-to-date until 1972/73 (Last,1978) . Wind direction dataar efro m 1974-1975.

and inth ehottes t month (August) from 17.8° Ca tnigh tt o32. 8° Ci nth e day. Maximum temperatures reach sometimes values'o fove r4 0°C . The relative humidity iso nth eaverag e5 8 %, ranging froma monthl y lowo f 43 %i nMa yt oa maximu m of7 0 %i nJanuary . Total annual evaporation froma n open water surface amountst osom e 2500mm ,measure d ina Clas sA Evaporation Pan (Last, 1978, qf.Arzi , 1981). Thepotentia l annual evapotranspiration is 1917 mm (Arzi, 1981), calculated according toStanhil l (1961). Detailed monthly climatic dataar epresente d inTabl e7 ,modifie d from Arzi (1981).

64 HYDROLOGY The hillslope hydrology of asomewha t similar area as Horvat Haluqim, situated on the sameanticlina l range, has been studied indetai l by Yair (1981) and his collaborators. Their main conclusions from the Sede Boqer experiment site, some4 k m southwest ofHorva t Haluqim, about runoff genera­ tion and its flow down the hillslope can be summarized as follows: (1)Bedroc k areas respond more quickly to rainfall than do soil covered areas. (2) An integrated flow of runoff along the entire slope, that includes a colluvial part toward the bottom, isquit e rare under the present rainfall conditions in the area (Yair, 1983). On the hillslopes studied by Yair, three geological formations aresucces ­ sively exposed from top to bottom: the Nezer, Shivta,an d DrorimFormations . Deep colluvial deposits occur at the base ofth e slopes. Although the two sites appear ina similar landscape setting, there are some important diffe­ rences. The Drorim Formation, which appears to be less runoff productive than the other geological formations (Yair, 1981:242),doe sno tcro p out at the Horvat Haluqim site. Colluvial deposits at the base ofth e slopes are virtually absent at Horvat Haluqim and thus cannot obstruct runoff flows from reaching the wadi bed. Moreover, the upper part of the ShivtaFormation , which shows the best runoff production at the Sede Boqer experimental site, is exposed at an advantageous geomorphicpositio n atHorva t Haluqim: at the lower part of the hillslopes just above the three wadis. The shape ofth e hillslopes at Horvat Haluqim is usually convex, having amaximu m gradient of about 20-25 degrees, generally atthei r lower part toward thewadis .Averag e slope length from interfluve towad i bed isabou t 100m on thehill y spurs that seperate the three wadis from each other. Considering the outcome of the research conducted by Yair and his col­ leagues at the SedeBoqe r site, it seems that thethre e wadis of Horvat Haluqim are endowed with an advantageous runoff-producing and runoff- conducting catchment. Although the catchments are relatively small, obvious for stream channels ofth e first and second order, their geologic and geomor­ phic properties appear well suited for runoff farming.

SOILSAN D VEGETATION This preliminary conclusion seemssubstantiate d by thedens e wadi vegeta­ tion, which is in sharp contrast to the scattered shrubs on the rather barren hillslopes (Photo 1, p.30). The sedimentology and soil development of the

65 v M ~..,.:^.:.. . ••.:, ~?i%m0^# *^" **^"i..

PHOTO 10. View fromon eo fth eterrace d wadisa tHorva tHaluqim , overlooking the synclinal Zinplai n towardsth e elevated Eocene Avdatplateau .

PHOTO. 11. Hard limestone bedso fth eNeze rFormation ,easil y quarried.

66 alluvial wadi deposits at Horvat Haluqim were investigated in detail in one of the man-made terraces.Th e results are presented in a following sub-chapter. The soil cover on both theNeze r and ShivtaFormation s is limited, patchy and shallow. Bare rock outcrops occur frequently throughout the hilly catch­ ment of the three wadis. The shallow lithosols on thehill s are dominated by the Artemisia herba-alba -Gymnocarpo s decander and Hammadetum scopariae asso­ ciations, the latter being characteristic for the driest habitats with the worst moisture-retention regimes. Deeper brown lithosols with significant soil moisture amounts exist in patchy rock crevices near bare rock outcrops,a sa natural mini-variety of the micro-catchment system in runoff farming. TheVarthemi a iphionoides - Origanum dayi association occurs at these most favourable runoff receiving "mini infil­ tration basins"o n the hillslopes (Yair and Danin, 1980;Danin , 1983:40-41). The salinity (EC)o f lithosol A-horizons at theSed e Boqer site, investi­ gated by Arzi (1981), ranges from 0.4-3.9 mS/cm. The salinity usually increases somewhat inth e subsoil,i fther e is a subsoil atall . The serozemstha t developed in thecolluviu m at thebas e ofth e slopes are much more saline inth e deeper subsoil. The electrical conductivity (EC) reaches here amaximu m value of 26mS/c m (Arzi, 1981). An article about similar colluvial deposits at the Haluqim anticlinal range, which includes soil stratigraphic data and their paleoclimatic implications, is currently in press (Issare t al., 1986). TheLat e Quaternary Serozem soils developed in the loessial deposits of the Sede Zin syncline are also very saline. TheA-horizo n has an EC of about 1.4 mS/cm, whilst values of4 2mS/c m already occur at adept h of 25 cm. Gypsic horizons are also found inthes e soils.Youn g alluvial loess soils on the Sede Zin plain have a very low salt content (Dan et al.,1973 ;Dan , 1981).

ARCHAEOLOGY OF HORVAT HALUQIM

The site extends along the three mentioned tributary wadis ofNaha l Haroa (Figure 9). Archaeological excavations at the sitewer e conducted by Rudolph Cohen (1976). The remains consist ofa small fortress and a settlement which includes 25 buildings, 4 cisterns and related agricultural terraces. Apart from two buildings ascribed to the 2nd-3rd centuries A.D. (Roman period), most of the other remains are attributed by Cohen (1976)t o the 10th century B.C. (Israelite period).

67 FIGURE 9. Plano fth eHorva t Haluqim site (modified afterCohen ,1976 ,1981) , showing thelocatio no fth e fortress, the watch-tower, various buildings, cisterns, and the threewadi swit h their terraced fields (surveyed by the author).Th egeograph y ofth esit eca n be summarized asfollows :

ALONG THEEASTER N WADI: Fortress 23x 2 1m 40m N ofth eroa d 10m Eas t ofth ewad ibe d Building 8 x 8m 90m N Eo fth e fortress 25m Eas t ofth ewad ibe d Building 14x 9m 90m NN E ofth e fortress JustEas t ofth ewad ibe d Cistern 18x 8m 130m NN E ofth e fortress 10m Eas t of thewad ibe d Building 10x 10m 140m NN W ofth e fortress 45m Wes t ofth ewad ibe d

68 ALONG THE CENTRAL WADI: Several buildings 180m NW of the fortress Just West of the wadi bed

ALONG THE WESTERN WADI: Cistern 22 x 12m 420 m NW of the fortress 65 m East of the wadi bed Cistern 11 x 11m 440 m NW of the fortress 25 m East of the wadi bed Cistern 14 x 9m 540 m NW of the fortress Inside the wadi bed Building 10 x 8m 480 m NW of the fortress Just West of the wadi bed

Some of the 25 recorded buildings, belonging to the settlement,wer e exca­ vated by Cohen (1976). The fortress has a roughly oval plan (about 21-23 m in diameter) and includes 7 casemate rooms around a central courtyard (Photo12) . Only the foundations of the fortress walls of Horvat Haluqim were found to be preserved, resting on the limestone bedrock of the Nezer Formation. The walls are built from local, roughly-hewn limestone blocks.Th e size of the casemate rooms varies from 1.5-2 mi n width to 5.5-8 m in length (Cohen, 1976). The shape of its ground-plan is similar to the Early Fortress at Kadesh-Barnea and other fortresses in the region (Cohen, 1980, 1981, 1983).

"'^S-SKiSf's;- '- •"*; ? ','

''*I K

** (>Clfe ;*%;#

PHOTO 12.Th e western casemate room of the fortress at Horvat Haluqim.

69 Cohen (1980:77)admit s that "the fundamental problem of the early fortress network is its date".Radiocarbo n dates from the tell atKadesh-Barnea , partly published inChapte r 8, seem to suggest that the Early Fortress at the latter site may have been erected in theLat e Bronze orEarl y Iron age. The apparent similarity in architectural style and shape ofth e groundplan between the Early Fortress at Kadesh-Barnea and the fortress at Horvat Haluqim (Cohen, 1980)raise s the question whether the early fortress network might date to the Late Bronze or Early Ironage . Cohen (1980)mention s the different archaeolo­ gical opinions inthi smatter , by which Rothenberg (1972)seem s to favour a Late Bronze age for thecerami c complexes involved. Recent advances inth ecalibratio n of radiocarbon dates from conventional C-14year s intohistorica l years (Chapter 8; Pearson, 1986), may lead to a situation in which theoutcom e of controversial stratigraphic problems in archaeology, based onceramic s or architecture,wil l beresolve d by radiocar­ bon dating on the basis ofchrono-stratigraphy . The dimension of time is indeed the absolute basis in stratigraphy and the impartial arbitrator in all aspects ofstratigraphi c reserach. Only time allows for the juxtaposition of contemporaneous events in history, archaeology, climate,an d landscape. In the absence of such a detailed chrono-stratigraphy for Horvat Haluqim, the age of the sitewil l be referred to in accordance with the general opinion of Cohen (1976), as Iron age, bearing inmin d the raised question that some parts ofth e site may be older perhaps.Tw o buildings at the site are actually younger and belong toth eRoma n period. The Roman watch-tower, situated 140m NNW of the fortress, isth e best preserved building atHorva t Haluqim. The walls.ar ebuil t of large hewn stones and are 1m wide.

WATER SUPPLY IN THE PAST

The nearest spring in the vicinity, fed by a natural flow of shallow groundwater, isth ebrackis h spring ofEi n Mor in theZi n Valley, some 7 km south ofHorva t Haluqim. The water of this spring isal l right for animals, but is considered too saline for regular human consumption. The necessary water supply for the former inhabitants ofHorva t Haluqim, therefore, had to come from runoff rainwater,catche d and stored incisterns . The four cisterns located in the settlement are built of large rough stones. All cisterns, except the one located inside the wadi bed of the western stream valley, are silted up to the brim. Three large piles of silt

70 next to the latter cistern, which still may fillwit h water in the rainy winter season, indicate that itwa s cleared anumbe r oftime s in the past. This cistern isova l in shape (9x 14.5 m), with walls 1m wid e and preserved to aheigh t ofA m . Itsestimate d water-holding capacity amounts to some 300 cubic metres of runoffwate r (Cohen, 1976). As the other threecistern s are completely silted up,i t ismor e difficult to assess their water holding capacity. Their combined volumes are conjectured tob e some 1100cubi c metres. Ifal l the cisterns would have filled up during the rainy season, approximately 1400cubi cmetre so fwate r could have been stored within the settlement in the past. Ina drough t year, probably much less runoff water could be harvested and stored, perhaps less than athir d of the estimated storage capacity, e.g. some 300cubi c metres. How many people could have lived on such amounts? Assuming adail y water requirement of 10lite r per capita (Evenari et al., 1971; Broshi, 1980), which isa rather low quantity,30 0cubi c metres ofwate r might have supported theminima l needs of some 80people . An amount of 1400 cubic metres, possibly stored in wet years, might have been sufficient for a population of 380. An assessment ofpas t population levels must be based, however, apart from other considerations, upon minimal figures in dry years (Mayerson, 1967; Broshi, 1980), as itwa s difficult tobrin g in any signifi­ cant quantities ofwate r in antiquity. Moreover, thewate r needs of possible flocks and other animals in the village must also be taken into account, besides evaporation lossestha t can amount to2. 5 meter/year (Table 7), if the water surface of thecistern s was fully exposed. It seems,therefore ,tha t the past population ofHorva t Haluqim was probably less than 80people .

ANCIENT RUNOFF-FARMING WADI TERRACESA THORVA T HALUQIM

The terraced fields in the wadis are in all likelihood contempora­ neous with many of the other structures within the settlement. The siting ofa number of houses atHorva t Haluqim seems tohav e been chosen to facilitate easy access to the cultivated fields in the terraced wadis, as already noted by Cohen (1976). Ifno t for the practising of runoff farming,ther e appears to be no logical explanation for the location of these houses in the settlement. Horvat Haluqim seems tohav e been a small agricultural village with a defen­ sive posture,i fth e fortress iscontemporaneou s with the terraced fields. The number and dimensions of the check-damsan d terraced fields in the

71 three wadiswer e surveyed by theauthor . Beginning atth ehea d ofeac h stream valley,th e damsan d terraced fields in therespectiv ewadi swer e successively numbered,whil e going downstream. Iti stheoreticall y possible that aver y few damsar ehidde n fromview ,completel y buried by alluvial sediments.Thi swoul d not affect,however ,th edetermine d outcome ofbot h thecombine d length ofth e terraced fields and thecalculate d cultivable area ineac h respective wadi:

(1) TheEASTER N Wadi exhibits 14check-dam san d terraced fields, having a total cultivable area of0.5 3 hectare. Thewad i isterrace d over a length of25 0m . (2) TheCENTRA L Wadi contains3 0check-dam san d related terraced fields, having atota lcultivabl e area of0.7 5 hectare. Thiswad i isterrace d over a length of45 4m . (3) The WESTERNWad iexhibit s 20check-dam s and related terraced fields, having atota lcultivabl e area of0.7 3 hectare. Thewad i is terraced over a length of41 1m .

This latter figure includesa possibl e break interracin g over some6 0 m between check-dam 3an d4 . Iti sno t clearwhethe r this stretch ofwad i was never terraced because ofa n apparently steeper wadigradien t or that erosion hasremove d onceexistin g damsan dterraces . The Western Wadi containsa cister n init sstrea m bed just above the beginning ofth e first terraced field. Thisset-u pmigh t have enabled inten­ sive vegetablecultivation , even during the summer season, in the firstter ­ raced fieldsbelo w thecistern ,irrigate d by its stored runoffrainwaters .Th e very existence ofhouses , situated justbelo w the first three terraced fields (Cohen, 1981:60), seems tostrengthe n thishypothesi so f an intensive and perennial vegetable-herb-fruit garden near home, watered from thecister n in addition tonatura l runoffflows . A very small tributary of theWester n Wadi joinsth emai n stream bed some 90m below thementione d building remains, inbetwee nda m 8 and 9. The fact that eventhi s littlean d narrow wadiwa sterraced ,show s the determination of the ancient villagepopulatio n tocultivat e every littleplo t of soil suited for runoff farming. The wadi has6 small check-damsan d related terraced fieldswit h atota l cultivatable areao f0.0 5 hectare (497squar e metres). The littlewad i isterrace d over a length of5 6m . The total area ofterrace d fields in thethre ewadi samount st o2.0 7 ha, whilst thetota lcatchmen t area ofth e threewadi scombine d is estimated at

72 about 36ha . The ratioo fth e runoff receiving areao fcultivate d fields to the runoff producing catchment area can, therefore, bepu ta tapproximatel y 1:17 forth e Horvat Haluqim site. This local datumi sver y nearth e average figure of1:1 8 forth e entire RunoffFarmin g Districti nth e central Negev desert (Kedar, 1967).

TABLE 8. Numberan dsiz eo fdam san dterrace d fieldsi nth e three wadis* o f Horvat Haluqim, descending from the heado fth e valleys downstream.

TERRACED DAMLENGT H LENGTH(m )O F FIELDARE A FIELD =WIDT H(m ) OF TERRACEDFIEL D (squaremetre ) and TERRACEDFIEL D (upstreamfro mdam ) DAM wadi wadi wadi wadi wadi wadi wadi wadi wadi NUMBER WEST CENTRAL EAST WEST CENTRAL EAST WEST CENTRAL EAST

1 15.5 12.5 12.0 30.5 12.0 12.0 398 132 160 2 24.0 13.0 14.7 28.4 16.8 16.8 560 214 224 3 18.0 15.2 15.7 15.4 12.0 18.0 323 169 274 4 10.0 18.0 19.3 (72.0) 17.3 38.7 224 287 677 5 13.0 19.4 18.6 12.5 31.0 14.4 143 579 273 6 14.5 13.0 20.0 12.6 5.6 9.6 173 33 185 7 11.0 16.0 19.6 13.7 21.0 9.6 174 304 190 8 10.0 17.5 20.0 12.3 16.0 15.7 129 268 311 9 28.5 15.6 22.0 17.4 15.6 13.7 335 258 288 10 32.5 15.7 24.0 14,2 13.0 14.7 433 203 338 11 31.5 11.5 24.1 14.6 21.3 18.3 467 289 440 12 29.0 13.5 23.5 40.0 14.5 20.0 1210 181 476 13 24.0 13.5 38.0 16.8 23.0 22.3 445 310 686 14 32.0 17.5 25.0 20.0 18.2 26.0 560 282 819 15 25.8 20.0 17.5 11.2 505 210 16 24.3 20.0 27.0 12.2 616 244 17 13.0 20.0 17.0 10.3 3Y? 206 18 9.0 19.3 7.2 10.1 79 198 19 8.0 15.9 9.5 12.0 80 211 20 10.0 15.7 12.2 18.8 109 297 21 15.8 11.7 184 22 23.0 19.2 372 23 26.7 17.7 439 24 26.4 14.8 392 25 22.5 20.5 501 26 17.0 13.0 256 27 13.3 9.8 148 28 11.2 7.5 91 29 9.5 13.0 134 30 10.3 15.2 150

The field widtho fa certai n terrace, usedt ocalculat eit sfiel d area,i s based uponth eaverag e lengtho fth etw o dams bounding that terraced field. * Data aboutth elittl e tributaryo fth e Western Wadi appeari nTabl e9 .

73 TABLE9 .Numbe ran dsiz eo fdam san dterrace d fieldsi nth elittl e tributaryo fth e Western Wadio f HorvatHaluqi m (measured from the heado fth e valley downstream).

TERRACED FIELD DAMLENGT H LENGTH (m)O F FIELDARE A ANDDA M number =WIDT H(m )O F TERRACEDFIEL D (metre TERRACEDFIEL D upstream fromda m square)*

1 6.0 6.0 30 2 5.5 7.5 43 3 5.5 9.0 49 4 8.0 7.8 52 5 13.0 12.2 128 6 17.0 13.0 195

The field widtho fa certai n terrace,use dt ocalculat eth efiel d areao ftha t terrace, isgenerall y based upon the average length ofth etw o dams bounding that terraced field.

The average ratio betweenth elengt ho fa terrace d fieldan dit swidt h (= the length ofth eda macros sth e wadi)i nth e runoff-farming terraced wadi system at Horvat Haluqimi s0.90 . Thusth e average lengtho f the terraced fieldsi sslightl y lesstha nth e average widtho fth ewad i bed. However, the individual terrace length/width ratio'sa tth e site vary considerably and range grosso modoi nbetwee n0. 5an d2.0 . Fora nequa l distributiono frunof f water within each terraced field,i t is important thatth efield sar elevelled . Ina prope r designo fa terrace d wadi system, the gradiento fa wadi , therefore,determine sho w oftenan da twhic h points check-dams aret ob econstructe d across thewad i bed. Iti s obvious thatth esteepe r thegradien to fth ewad i themor e check-damsan dterrace sar e requiredpe runi to fwadi-lengt ht oenabl e each terraced fieldt ob elevelled , in analogyt oa staircase . Apart from the wadi gradient, the deptho f soil within thewad ibe di sa determinin g factora swell .A relativel y shallow soil will not allow much roomfo rlevelling . Inth ecas eo ftw o similar wadis, having anequa l gradientbu tdifferen t overall soil depth, more check-dams might be necessary inth e wadi withth eshallowe r soil. Itremain s tob e investigatedho wleve lth eancien t terraces actually were. The three wadiso fHorva t Haluqim, includingth elittl e tributaryo fth e Western Wadi, containa tota lo fsom e7 2terrace d fields.Th e combined length of all the measured damspu ttogethe r amountst o123 4meter , whichi squit ea

74 constructional achievement for such a little village in order to wrestle a living from the desert.

TABLE 10. Data about runoff farming fields at theHorva t Haluqim site. (Terraced wadi system)

TRIBU­ WADI WADI WADI TOTAL OR TARY WEST CENTRAL EAST AVERAGE

a)Numbe r of dams and 6 20 30 14 72 terraced fields b)Averag e length of 9.3 17.5 15.1 17.8 15.4 (m) terraced field c)Averag e dam length 9.2 19.2 16.6 21.2 17.1 (m) = field width d)Rati o b/c =Rati o e/f 1.01 0.91 0.91 0.84 0.90

e)Tota l length of 55.5 350.8 454.3 249.8 1110.4 (m) terraced wadi fields f)Tota l length of dams 55.0 383.6 498.5 296.5 1233.6 (m) g)Tota l area of fields 497.0 728A 7542 5340 20663 (m2)

STRATIGRAPHY AND SOIL DEVELOPMENT OF AWAD I TERRACE AT HORVAT HALUQIM, CONSTRUCTED AND USED FOR RUNOFF FARMING IN ANTIQUITY

THE DISCOVERY OF AN ANCIENT TERRACE LAYER InNovembe r 1982a n excavation was made by the author into a terrace of the Eastern Wadi, some 150m north of the fortress,jus t upstream from dam 12nea r the Roman watchtower (Figure 9, Table 8). The dam has a length of 23.5 m and awidt h of 1.60 m. It is built across thewad i ina southwest-northeast direction, its azimuth being 62degrees .Althoug h the architecture of the dam was not investigated, its rather complex and somewhat dilapidated appearance isunlik e the precise and neatly layered dams, so often characteristic for the Nabatean-Roman-Byzantine period. It ispossibl e that the dam was first built during the Israelite period, in accordance with the age attributed to the site by Cohen (1976). New evidence for the pre-Nabatean origin ofthes e agricultural terraces was found during the present excavations, as described below. Independent carbon- 14 dating may bepossibl e in a later stage. It seems likely that the dam may

75 •*&>*i^> rTx?--.. >'•

PHOTO 13.Excavatio n into terrace 12 of the Eastern wadi of Horvat Haluqim.

also have been repaired and expanded for runoff farming in later periods, as indicated by some Roman remains at the site. Thus the dam probably reflects a composite archaeological and architectural history. The terrace upstream from the dam is slightly affected by erosion. A very shallow but discernible erosion channel winds its way through the terrace, and has made an opening into the somewhat dilapidated terrace dam described above. The first pit into the terrace was excavated from 0-250c m upstream from the dam, just east from the very shallow present stream channel. The location of this pit proved to be well chosen, as, rather surprisingly, a stone wall was discovered below the surface at a depth of 50-100 cm. This wall begins at the contact with the dam, and runs perpendicular from the dam in an upstream direction in the middle of the terraced wadi bed. The wall is built of two courses, being only one stone wide and about 50 cm high. Local natural limestone blocks were used in the construction of this wall, which might have been part of an ancient spillway, buried by alluvial loessial sediments after the site was abandoned. The wall has so far been excavated until A m upstream from the dam. Its upper course consists of 12 stones.

76 PHOTO 14. Exhumed wallan dforme r terrace surface, Inth emiddl eo fth e wadi bed,perhap s dating backt oth eIsraelit e period (Ironage) .

^ i4 t ' .Ska- ^^hJ iWl - &» JT._• - -»!.-*.•

50

-CD lll-A 65 Pit-2 C3> C3QCS 85

lll-AQ 100 a 115 125 130 135 %; Carbonate nodules o o o 140 150 o°oo Gravel IV-C * 155 * * 160 Limestone 165 Block ~3 5 °"~ 175 CD o o 185 ° O ° o V-C o » • o O 1

FIGURE 10. Excavatedan dsample d soil pitsi nth e12t h terraceo fth e Eastern wadio fHorva t Haluqim (relatedt oPhot o14) .

77 The average length of the limestone blocks making upth e wall is some 33 cm, similar to theirwidth ,whils t their thickness (height)i sabou t 25 cm. Natural blocks of limestone were found on both sides of the wall at about the samedepth , sometimes appearing to form aprotectiv e cover forth e under­ lying loessial sediment, that could otherwise have been affected by erosion and piping, as a result of runoff flows, by which the dam might have been undercut. If this interpretation iscorrect , then these horizontally spread limestone blocks are likely to have formed part of an ancient spillway system, together with the two-course high wall. However, since this wall has not yet been excavated over itsentir e length,th e above conclusion remains tentative. The alluvial loessial soil shows a somewhat different character from about 50-100 cm, at thesam e depth as the buried wall. Thecolou r becomes more greyish, the structure more compact, and a few black and brick-red spots appear occasionally. A few tiny pieces of bone were also discovered. This is evidently a layer disturbed by man inth e past, a buried surface horizon. Pottery sherds found inthi s layer are attributed by Cohen (1983, personal communication) to the Israelite period. It thus seems clear that,approximate ­ ly in themiddl e ofth ewad i bed, a pre-Nabatean, possibly Iron age soil surface has been discovered in relation to the two-course high wall and adja­ cent rough pavement. Thewal l is built on alluvial loessial sediments, of a silt loam texture, which below a depth ofabou t 100c m do not appear to have been disturbed by man. Small mounts of gravel and a few stones occur inth e fine-grained allu­ vial loessial deposits. At a depth of 140-170 cm aweakl y developed calcic horizon occurs, showing small but distinct orthic carbonate nodules (Wieder, 1977; Wieder and Yaalon, 1974), making up about 1-3 volume per cent of the soil matrix.Belo w adept h of 175c m the amount of gravel increases considera­ bly,bu t at 225 cm depth the bedrock surface has still not been reached. A second pit was excavated unto adept h of7 5 cm, a fewmete r east ofth e former one. The silt loam loessial profile is rather uniform in its appea­ rance. Apart from a few small sherds of pottery of doubtful origin, no archaeological objects of stratigraphic value were found in this pit,althoug h some natural limestone blocks appeared scattered at about the same depth as the upper part of the discovered wall in pit-1 (Photo 14).Th e second pit lies somewhat closer to the hillside edge of the terrace,an d farther away from the very shallow erosion channel.Th e surface of the terraced field is slightly higher near pit-2tha n around pit-1,a serosio n has apparently removed part of the surface layer near the latter pit.

78 LABORATORY DATA Soil samples were collected from both pits. The samples inpit- 1 were taken some 70 cm west of the wall discovered in the subsoil, and 245 cm upstream from the dam.Th e samples in pit-2wer e taken at adistanc e of 250c m east of the buried wall, and 80 cm upstream from the dam (Fig. 10).Th e actual distance between the sampled profiles is35 0cm . The depth of the samples is stated with respect toth e adjacent surface ofth e terrain next to the two pits,whic h appears tob e somewhat lower near pit-1. The content of organic carbon and nitrogen can be rated as very low (Ilaco, 1981) for the entire alluvial soil profile of this ancient runoff-farming terrace,whic h ischaracteristi c for arid regions.Despit e thesedeficiencies , negatively affecting natural soil fertility, the loessial soil is physically and structurally well suited for runoff-agriculture, having a good water- holding capacity. The most striking contrast between the twopit s isth e large difference in the respective contents of soluble salts. Pit-1 can beregarde d ashavin g an extremely low salinity content (Ilaco, 1981): itselectrica l conductivity ranges from 0.23-0.38mS/cm . The samples from pit-2, at adistanc e of only 3.50 m east of the sampled profile of pit-1, exhibit, on the other hand, a medium to moderately high salinity. The electrical conductivity gradually increases from 5.95 mS/cm in the upper 10c m to9.0 0 mS/cm at adept h of 70 cm. The proximity of pit-1 to the present, very shallow, erosion channel suggests that, (1)eithe r accumulated saltshav e been leached here by recent runoff stream flows from the entire profile beyond adept h of 1.85 m, or (2) no salt accumulation could ever take place on thisparticula r spot since the end of runoff farming at this terrace. The facttha t pit-2, at a sampling distance of only 3.50 m (!), doescontai n almost 30time smor e salts is quite extraordinary. The alluvial loessial sediments in pit-2see m for themos t part to post-date the Iron-Age. Thus the saltsmus t alsohav e accumulated since that period, after the final termination of runoff farming. The accumulation of salts in soilsma y require considerable time (Yaalon,1964 ;Da n and Yaalon, 1982), whereas leaching may occur within a fewyears . Since the original salt content of the alluvial deposits isunknown , it is hard to assess how long aperio d it took for this salinity profile todevelop . The much higher salinity of pit-2, ascompare d topi t 1, doesmak e itclear , however, that runoff flows have not been able to flood this part of the terraced field, which isonl y slightly higher than the area around pit-1, for

79 acertai nperio do ftim eunti ltoday .

TABLE 11.Granulometr yan dsoi lchemistr yo falluvia lloessia lsediment si na n ancientrunoff-farmin g terraceo fth eEaster nwad ia tHorva tHaluqi m 1)Uppe rsediment s- Pit-2.

HORIZON DEPTH CLAY SILT SAND TEXTURE GRAVEL CARBONATES <0.002 0.002- 0.063- >2m m CaCOj mm 0.063 2 mm (cm) (%) (%) (.%) (%) (X)

A 0-10 17.1 55.8 27. 1 SILTLOA M 2 30.0 C 10-20 16.4 60.7 23 0 SILTLOA M 5 26.9 C 20-30 14.9 62.4 22 7 SILTLOA M 2 30.6 II-C 30-40 20.2 58.8 22 0 SILTLOA M 8 30.3 II-C 40-50 19.3 51.7 29. 0 SILTLOA M 12 29.6 II-C 50-62 20.6 54.0 25 4 SILTLOA M 9 29.1 II-C 62-75 26.1 50.4 23 4 SILTLOA M 4 29.4

HORIZON DEPTH pH EC(1:1) SAR ORGANIC NITRO­ TOTAL CARBON GEN PHOSPHORUS (cm) (mS/cm) (%)c (%) N (X)P 205 A 0-10 7.4 5.95 47.9 0.56 0.05 0.23 C 10-20 7.4 7.59 77.1 0.56 n.d. 0.24 C 20-30 7.4 7.70 83.7 0.52 0.04 0.28 II-C 30-40 7.5 7.90 79.5 0.56 n.d. 0.35 II-C 40-50 7.4 7.99 66.7 0.53 n.d. 0.34 II-C 50-62 7.4 8.70 72.7 0.57 n.d. 0.33 II-C 62-75 7.3 9.00 74.9 0.58 n.d. 0.32

HORIZON DEPTH SOLUBLE SALTSI N 1:] SOIL/WATEREXTRAC T ( •neq/kg soil)

2+ >.2 + ..+ + 2 (cm) Ca Mg Na K HC03" CI" S04 " N03"

A 0-10 1.7 <0.1 0 52.0 0.8 0.9 53.0 0.4 0.4 C 10-20 1.6 <0.1 0 70.0 0.8 1.9 65.0 5.9 0 c 20-30 1.4 <0.1 0 70.0 0.8 0.8 65.0 5.4 0.2 II-C 30-40 1.5 <0.1 0 70.0 0.8 0.9 58.0 7.5 0 II-C 40-50 2.1 0 .10 70.0 0.5 0.8 55.0 13.5 <0.1 II-C 50-62 2.3 0 .15 80.5 0.8 0.7 68.0 14.0 <0.1 II-C 62-75 2.5 0 .20 87.0 0.8 0.6 75.0 16.0 0

Percentagesi nweigh t %o fth efin eeart h< 2mm ; onlygrave li nweigh tX o f the total soil. EC= electrica lconductivity , SAR= sodiu m adsorption ratio, n.d.= no tdetermined .

Sampling depth inbot hpit si swit hreferenc et oth esurroundin g surface, which seems to besomewha thighe rnea rpit-2 . Thisshoul d be taken into accountwhe nbot hpit sar ecompare d stratigraphically:i ntha tcas esom e1 5c m mustprobabl yb eadde dt oth esampl edepth so fpit- 1(Tabl e12) .

80 TABLE 12.Granulometr yan dsoi l chemistry ofalluvia l loessial sedimentsi na n ancient runoff farming terraceo fth eEaster n Wadia tHorva t Haluqim 2)Burie d Iron-Age soil, overlying Late Quaternary alluvium -Pit-1 .

HORIZON DEPTH CLAY SILT SAND TEXTURE GRAVEL ;ARBONATES <0.002 0.002- 0.063- >2m m CaC03 mm 0.063 2 mm (cm) (%) (%) (%) (%) (%)

III-A 50-6 5 14.7 50.0 35.3 SILTLOA M 6 30.9 III-A 65-8 5 13.1 58.2 28.7 SILTLOA M 7 33.1 III-A(C) 85-100 14.9 63.6 21.5 SILTLOA M 5 29.4 III-A(C) 115-125 18.0 62.3 19.7 SILTLOA M 5 31.3 IV-C 130-135 12.2 61.9 25.9 SILTLOA M 6 30.1 IV-Cca 140-150 17.8 56.6 25.6 SILTLOA M 11 30.4 IV-Cca 150-155 17.9 57.5 24.6 SILTLOA M 7 31.5 IV-Cca 160-165 17.3 67.7 15.1 SILTLOA M 6 29.3 V-C 175-185 20.1 61.1 18.7 SILTLOAM/GRAVE L 17 29.6

HORIZON DEPTH pH EC(1:1) SAR ORGANIC NITRO­ TOTAL CARBON GEN PHOSPHORUS (cm) (mS/cm) (%)c (%) N (%)P 2O5 III-A 50-6 5 7.8 0.34 3.2 0.69 0.04 0.55 III-A 65-8 5 7.9 0.23 4.5 0.74 n.d. 0.52 III-A(C) 85-100 7.8 0.26 5.6 0.46 n.d. 0.36 III-A(C) 115-125 7.8 0.30 8.1 0.65 n.d. 0.40 IV-C 130-135 7.7 0.25 6.3 0.35 n.d. 0.13 IV-Cca 140-150 7.7 0.33 9.6 0.23 n.d. 0.14 IV-Cca 150-155 7.7 0.33 8.9 0.23 n.d. 0.13 IV-Cca 160-165 7.7 0.37 12.8 0.42 n.d. 0.15 V-C 175-185 7.6 0.38 14.1 0.46 n.d. 0.13

HORIZON DEPTH SOLUBLE 5ALTSI N 1:1SOIL/WATE REXTRAC T (meq/kgsoil )

2+ 2+ + + 2 (cm) Ca Mg Na K HC03 CI" SO N03

III-A 50-6 5 0.5 <0 .10 1 .8 <0 1 2.1 0.5 0 .2 <0.1 III-A 65-8 5 0.5 <0 .10 2 5 <0. 1 2.4 0.2 0 4 0.2 III-A(C) 85-100 0.4 <0 .10 2 .8 <0 1 2.6 0.2 0 3 <0.1 III-A(C) 115-125 0.3 <0 .10 3 2 0. 2 2.9 0.3 0 4 <0.1 IV-C 130-135 0.3 <0 .01 2 .5 0 2 2.2 0.4 0 .2 0 IV-Cca 140-150 0.3 <0 .10 3 7 0. 6 3.5 0.5 0 2 0 IV-Cca 150-155 0.2 <0 .10 3 .2 0 4 3.3 0.3 0 .4 0 IV-Cca 160-165 0.1 <0 .10 3 .5 0. 4 3.5 0.2 0 3 0 V-C 175-185 <0.1 <0 .10 3 .2 0 3 3.5 0.2 0 .4 0

Percentages inweigh t% o fth efin e earth< 2mm ; only gravel inweigh t %o f the total soil. EC= electrica l conductivity. SAR =sodiu m adsorption ratio, n.d.= no tdetermined .

81 SOIL MICROMORPHOLOGY Microscopic studies of thin sections from the buried Iron-Age soil layer of Horvat Haluqim, showed unequivocally that wet conditions have existed in the past, presumably asa result of runoff farming. This conclusion isbase d on the occurrence of ferric nodules or concentrations of iron-(hydro)oxides in the Iron age soil horizon (Photos 15an d 16). Ferric nodules can only develop under an alternating wet-dry reduction-oxidation regime, which is quite abnormal for aridic soils in dry regions. These observed gley features seem to be convincing evidence of runoff farming practices in the past, in the time before the silting up of the terrace and the subsequent salinization. Runoff farming must have been successful in the sense, that the built terrace dam was able to keep a sub­ stantial amount ofrunof f water on this particular terraced field, during occasional flooding inth e rainy season, which caused a temporary reduction environment in the soil.

PHOTO 15. Dark ferric nodule in the centre of thephoto , with a diameter of about 0.25 mm, situated in the ancient terrace layer (probably Iron age),a ta depth of6 0 cm. Note a land- shell remnant at the right-hand side. This microscope-photo was taken with plain polarized transmitted light (without crossed polarizers).Th e length of the picture equals0. 5 mm. 82 Iron, which isusuall y present in the alluvial loessial soils under con­ sideration,ca n becomemobil e in awet ,reduce d environment.Throug h diffusion processes the ironma y move to certain micro-sinks inth e soil, near spots where voids facilitate the entry of air (oxygen). Here the iron is again oxidized and immobilized asa result. These diffusional micro-sinks may develop into concentrations of iron(hydro)oxides or ferric nodules after a certain period oftime . The formation of ferric nodulesma y take place within a relatively short time range in the order of one hundred to a few hundred years, depending upon various circumstances (Brinkman, 1986,persona l communication).Fo r moresoil - chemistry information on this issue,th e reader isreferre d toBrinkma n (1979) and Van Breemen (1986). Amor e detailed account of the soilmicromorpholg y of this terraced field atHorva tHaluqi m will bepublishe d elsewhere.

PHOTO 16. Dark ferric nodule in the centre of the photo, with a diameter of about 0.35 mm, situated inth e ancient terrace layer (probably Iron age),a ta depth of 75cm . Thismicroscope-phot o was taken with plain polarized transmit­ ted light (without crossed polarizers). The length ofth e photo equals 0.5 mm.

83 STRATIGRAPHIC CONCLUSIONS On the basis of the field and laboratory data a fivefold stratigraphic division is made within the investigated deposits, arranged in a soil- sedimentary sequence from the terrace surface downward. The stratigraphic indices are similar to those appearing inth e soil horizon designation of Tables 11 and12 .

I. POST IRON-AGE SEDIMENTATION PHASE-2. The upper 30c m of alluvial loes- sial sediment (pit-2) has an average clay content of 16.1 %, whereas the amount of total phosphorus (0.25% P 205)i slowe r than in the subsoil. The phosphorus content, very stable in acalcareou s environment, isunderstoo d to reflect the influence ofhuma n activity.

II. POST IRON-AGE SEDIMENTATION PHASE-1 (RUNOFF-FARMING PERIOD-2?). The alluvial loessial deposits at adept h from 30-75 cm (pit-2)contai n on the average more clay (21.6% )tha n the overlying layer, whilst the phosphorus content suddenly rises to about 0.34 X. Itseem s likely that the lower part of this layer istransitiona l to the Iron Age soil,exhibite d in pit-1.

III. IRON-AGELAYE R (RUNOFF-FARMING PERIOD-1).Th e buried Iron-Age horizon appears inpit- 1 at adept h from about 50-100 cm. The lower clay content of the alluvial loessial sediment may be due toth eclose r proximity ofpit- 1 to the main part ofth e stream channel. The phosphorus and organic C content reaches peak values in this horizon of0.5 5 %an d 0.74 %, respectively.

IV. PRE-TERRACING (PRE-RUNOFF-FARMING)ALLUVIA LLOESSIA L SEDIMENTS. Below the buried Iron-Age horizon, the phosphorus and organic C contents show a significant decline to0.1 3 %an d some 0.30 %, respectively. This clearly indicates the beginning of an alluvial loessial layer,predatin g the period of runoff farming and the influence ofman . Aweakl y developed calcic horizon (140-175 cm depth)i spresen t inthi s layer.

V.GRAVELL Y ALLUVIAL SEDIMENTS.Th e amount of gravel increases sharply at a depth of 175cm ,whic h marks the upper boundary ofcoarse r alluvial sediments, mixed with silt loam.A t 225c m depth,bedroc k was not yet reached.

A number of significant stratigraphic conclusions can be drawn from this excavation in the 12th terraced field of theEaster n Wadi ofHorva t Haluqim.

84 Since the IronAg e soil surface appears at an average depth of about 70cm ,i t seems beyond any doubt that loessial soil already existed in this wadi, when the terraced wadi system was first built forrunof f farming purposes. More than 155 cm of soil wasalread y present at this particular spot, when the ancients decided to terrace the wadi. The upper 100c m consisted of fine loessial soil, whilst the gravel content increased below this level.Th e more gravelly subsoil, alsomixe d with silt loam, was still capable of keeping moisture in itsprofile . Thus the entire 155+ cm of soil, already present at the onset of the ancient runoff-farming period, could have stored runoff water to sustain field crops or trees. Kedar (1957) expressed the view that oneo f the main purposes of the ancients to built theman y thousands of dams across thewadi s in the Negev was to enable soil to accumulate, so that agriculture could be practised in a later stage. Kedar apparently assumed that no soil orno t enough soil was present inman y wadis designated for runoff farming by the ancients,whe n they first decided to built these dams. This view clearly does not seem to fit the stratigraphy ofth eterrace d field described above. Enough soil was already present, when theancients , presumably the ancient Israelites (Cohen, 1976), built dams across the little wadi(s)a t Horvat Haluqim. The main and probably sole purpose to terrace thewadi swa s the retention of runoff water rather than soil. The fact thatmor e sediment also began to accumulate as a result of damming, was a secondary factor,no t always welcome perhaps, as it necessitated the building ofne w layerso fstone s on top of the dams in the course oftime .O n the other hand,th e annual addition of some new sediment with every flood onto the wadi terracesmigh t have contributed to rejuvenate soil fertility and keep it at ahighe r natural level than without this seasonal accumulation of eroded surface material from the slopes of the surrounding catchment.

85 Estimated wheat yields under runoff-farming conditionsi nantiquity ,relate dt oHorva t Haluqim

INTRODUCTION

How many people might have beenfe da tHorva t Haluqim inth e past with locally produced food, growni nth e three small terraced wadis? Iti s impossible to give anaccurat e answert otha t important question without modern research data from the actual site about yields of food crops in relation to year-to-year climate, crop rotation systems and manuring practices. Inmoder n research about runoff farmingi nth eNege v (Evenarie tal. ,1971 , 1982)th e Terraced WadiSyste m hasno tbee n investigated, whilst agricultural examinations of the Hillside Conduit System (Avdat and Shivta), Diversion System (Avdat)an dLima n System (Wadi Mashash)wer e largely concentrated on fruit trees.Som e data about wheat yieldsar e available from the Avdat experi­ mental runoff farmfo r the period from 1961/62t o1966/67 . These data,thoug h quite interesting, areo flimite d valuet oasses s yieldsi nantiquity , since chemical wereused , exceptfo rth efirs t year1961/62 . An initial beginning has been madeb yth e authort oinvestigat eth ecrucia l question of climatic variability indr yregion s and self-sufficient food production basedo nrunof f farming (Bruinse tal. , 1986b).Futur e researcho n this subject should involve systematic year-to-year food crop production, evaluated inrelatio nt oclimate , rainfallan drunof f amounts,cro p rotation, cropping patterns, andsoi l fertility (manuring). Such knowledgei svita lt o assess theviabilit yo frunof f farmingi nth e Negeva sa food-cro p producing system, inpas tan dpresent . Researcho nthi s issuema yhel pt oresolv e some

86 historic-archaeologic related questions,an dwil l provide know-how onclimate - defensive measures that might betake n with regard to runoff farming in developing countries. Despiteal lth eshortcoming s inou rknowledg e today,a roug h estimationca n bemad e about themagnitud eo fpas t food production atHorva tHaluqim ,a swel l asth eamoun to fpeopl e that might have subsisted onthi s local produce.

WHEAT YIELDSBASE DO NANCIEN T LITERARY SOURCES

About 73 % of theaverag e food basket today is composed of cereals (Buringh, 1977). Thispercentag e wasprobabl y even higher inth epast , when cerealsan dlegume s (pulses)mad eu pth emajo r parto fth ediet .Noteworth yi s the remark madeb yPlin y (Nat. Hist. 18.21.94-95; Loeb ed., vol5 ,p.249) : "Nothing ismor e prolific than wheat -Natur e having giveni tthi s attribute, because ituse dt ob ehe rprincipa l meanso fnourishin g man". Itthu s seems reasonable to baseth efollowin g calculations entirely oncerea l production and consumption data. The unique Nizzana papyrio fth e6t han d7t hcentur y A.D. are the only historical documents that provide actual datao nancien t yieldso ffoo d crops in theRunof fFarmin g District ofth eNegev . Document 82 (Mayerson, 1955; Kraemer, 1958) mentionsth eyield so fwheat , barley, andaracu s (alegume ) during acertai n yeari nth e7t hcentur y inth eNizzan a area. Apart from the fact that aracusi sa legume , itsspecifi c typological meaning is unknown

TABLE 13. Yieldso fwheat , barley andaracu si nth eNizzan a area,grow ni na certain year during the7t hcentur y A.D. under runoff farming conditions (after Mayerson, 1955).

PLACE CROP ModiiSOW N ModiiREAPE D YIELD

Ragoriost o Kat Wheat 40 270 6.8 -fold Malalkani Wheat 40 288 7.2 ,, Alphag Wheat 180 1225 6.8 ,, Alphag Barley 50 402 8.0 ,, Berain Barley 40 350 8.8 ,, Berain Aracus 30 97 3.2 ,,

87 (Mayerson, 1955). Yieldsar eusuall y stated inancien t literary sourcesa sth e volume amount harvested with respect toth evolum e amount sown. Theyield si n the Nizzana papyri areexpresse d interm so fmodi i sown and modii reaped (Table 13).On emodiu si s8. 7 litres (Mayerson, 1955:271). Today crop yieldsar egenerall y presented asth eweigh t of the harvested crop per unit of field area, often ink g or ton perhectare . Whereasa liter of water is alwaysa kg , a liter ofwhea twil lno t always have the same weight, asthi sdepend so n thesiz e ofth e grainsan d their specific weight. Variations in theweigh t ofa liter ofwhea t are, therefore, likely between different wheat varieties. A liter ofbarle y or aracuswil l likewise have different respectiveweights . Thusw e are faced with aproble m to translate ancient yields, expressed involumes , accurately intokg ,o r intokg/ha ,fo r comparison with presentyields . Mayerson (1955:55) estimates4 modi i tob e theapproximat e equivalent of oneU.S . bushel, whichmean stha t inhi sopinio non emodi io fwhea t isabou t 6.8 kg. In anarticl e onwheat-farmin g during Roman times, White (1963) appearst ohav e used the following equation: 1modi i= 5.8 8 kg ofwheat .Thi s latter figure isuse d inth e following calculations, to facilitatecompariso n withmoder n yields.Fe w other literary data exist aboutyield s in antiquity in theMediterranea n region.Whea t isusuall y theonl y crop onwhic h some ancient information isavailabl e (White,1963 ;Feliks ,1963 ;Sperber , 1978). The question arisesho wmuc h seed ofwhea twa ssow npe r unit of area in antiquity? Information on thismatte r may enable thetranslatio n of ancient sowing amountsan d yields into kg/hectare.I nadditio n it isimportan t tokno w what the ancientsthough t about sowing practiceswit h respect toclimati c and soilconditions .Th eNizzan apapyr i dono t contain information on thismatter . White (1963) studied both ofthes e aspects from ancient Roman literary sources. Despiteth e facttha t theavailabl eRoma n sourcesd ono t touch upon runoff farming assuch , their data arequit e instructive andma y serve as a basist oestimat e the likely amount ofwhea t sownunde r rather dryconditions . Therepor t by Cicero (InVerr. , II, iii, 112)abou twhea t yields inth e Leontini district ofSicil y in theyea r 70B.C . ismos t useful. Cicero does not only report theancien t wheat yields intha t region in theusua l volume­ tric comparativeway , asbein g 8-t o 10-fold theamoun t sown,bu t also states the amount ofwhea t sownpe r unit ofarea . This important historical text enabled White (1963)t ocalculat e theyiel d reported by Cicero as4 8modi i per iugerum. This iseithe r an8-fol d or a10 - fold yield according towhethe r the seed issow n atth erat e of6 or 5 modii

88 per iugerum. An iugerum is0.25 2hectar e (Mayerson, 1955:271). White (1963) translated this past wheat yield inSicil y of the 1stcentur y B.C. into the more modern expression of 11.2 quintals per hectare, which is 1120 kg/ha. Hence onemodi io fwhea t isestimate d to be 5.88 kg, whereby one modii per iugerum equals 23.33kg/ha . The amount ofwhea t sownpe r hectare in Sicily during the 1st century B.C. seems, therefore, to have ranged in between 112 and 140kg/ha .

THE QUANTITY OF WHEAT SOWN INRELATIO N TOSOI L AND CLIMATE

White (1963) reports the following information from ancient literary sources about variations of the quantity of seed sown per hectare with respect to soil and climate conditions inRoma n times: Varro merely recommends the farmer to adapt the quantity of seed sown to local soil conditions. Columella (DeR eRust. , II, IX,5-6 )provide s amor e comprehensive account ofth ematter :

"If the field ismoderatel y stiff (cretosus)o rwe t (uliginosus)yo u need for winter wheat (siligo)o r common breadwheat (triticum)rathe r more than five modii... But if the ground isdr y (siccus) and loose in texture (resolutus), no matter whether it be rich or lean,onl y four; for conversely, lean land requires the same amount of seed; unless it is sown thinly, it produces asmal l and empty head. But when it branches out into several stalks from one seed itmake s aheav y stand from a light sowing."

From this passage it isclear , as pointed out by White (1963), that the dominant consideration in deciding whether to sow thinly or thickly is the moisture content and texture of the soil. The same distinction is made by Pliny (Nat.Hist .XVIII ,199) :

"In dense, chalky ormois t soil sow six modii ofwinte r wheat or common wheat, but in loose, dry and fertile soil, four: Ameagr e soil produces a small and empty ear unless it has the stalks far apart, whereas fieldswit h a rich soil produce anumbe r of stalks from a single seed and yield aheav y crop from thinly-sown seed."

It was customary inGreec e and Italy to sow either early in autumn (light

89 rains)o r immediately before the onset ofwinte r (White, 1963)wit h itsheav y rains. Inearl y autumn the seed would be sown thinly, whilst just before the winter itwa s sown thickly (Theophrastus, Caus. pi., Ill, 25; qf. White, 1963). Columella (DeR eRust. , II, IX, 6)point sou ttha t the thin early sowing enablesth eplan t to form stoolsdurin g thewinte r and branchout . Inter-cropping ofcereal swit h olivesan d vineswa sextensivel y practised inRoma n Italy (White, 1963). It isno t knownwhethe r inter-cropping ofthi s kind was also carried out inth e arid Negev desert under runoff farming conditions in antiquity.Fro m the Nizzana papyri athreefol d land division can be deduced in the7t h century A.D. (Mayerson, 1955;Kraemer , 1958):

(1)Seedland ,mainl y used for food crops likewhea t andbarley . (2)Vineyard . (3)Garde n or orchard.

This differentiation seemst o suggest adivisio n ofcroppin g in the central Negev desert rather than theexampl e of inter-cultivation inRoma n Italy.

ESTIMATED AMOUNTO FWHEA TSOW N INTH ECENTRA LNEGE V DESERT ANDCORRESPONDIN G YIELDS

Considering theadvic e ofColumell a (DeR eRust. , II,IX, 5-6 ) andPlin y (Nat. Hist. XVIII, 199)t o sow 4modi i ofwheat ,presuambl y per iugerum,i n dry soil, it isassume d that asimila r amount of4 modii/iugerum , which is about 93.3kg/ha , wassow n inantiquit y in thedr y Negev desert. Thisi son e modii lesstha nth eamoun t of5 modi i per iugerum, assuggeste d by Mayerson (1955). The amount of93. 3k g seed per hectare fitsver y wellwithi n the range advised by the agricultural compendium for ruraldevelopmen t in the and (Ilaco, 1981), put at 70-100kg/h a for line sowing, and 25% more where broadcasting ispractised ,i.e .88-12 5kg/ha . The number of hectaressow n and the wheat yields inkg/h a can thus be calculated from document 82o f theNizzan a papyri, fora certai n year in the 7th century A.D. inth eNizzan a area, grown underrunof f farming conditions (Table 14).Th e averageyiel d amounts to64 6kg/ha ,whic h isconsiderabl y less than the figure of94 3kg/h a calculated by Mayerson (1955). The difference is caused by hisassumptio n of a sowing rate of5 modi i per iugerum, and one modii ofwhea t being equal to6. 8 kg.

90 TABLE 14.Calculate d yields ofwhea t and the amount ofhectare s sown in three individual historic cases inth e Nizzana area (central Negev)* , compared with past yields from selected other regions.#

PLACE TIME Modii Modii HECTARES YIELD SOWN REAPED sown (kg/ha)

Nizzana area 7thcent .A.D . 40 270 2.52 630 Nizzana area 7thcent .A.D . 40 288 2.52 672 Nizzana area 7th cent.A.D . 180 1225 11.34 635

Roman Sicily 70B.C . 1120 Sicily 1959 1060 Italy Varro (+3 0 B.C.) 1200 Italy Columella (+ 5 0 A.D.) 500 Syria 1949-1957 700 1949-1957 500 Israel 1910-1930 531

* Based on document 82o fth eNizzan a papyri (Mayerson,1955 ;Kraemer , 1958) # and data from White (1963)an d Stanhill (1978). The assumed sowing rate is 4 modii/iugerum (93.3 kg/ha), whereby 1modi i ofwhea t isassume d to have a weight of 5.88 kg.

EVALUATION OFANCIEN T WHEAT YIELDS INTH ELIGH T OFMODER N RESEARCH

The calculated averagewhea t yield of64 6kg/hectar e from the Nizzana area isslightl y above theoutcom e ofmoder n field experimentation of wheat yields in Israel from unfertilized plots, being onth eaverag e 580kg/h a (J. Ephrat, 1977,persona l communication, qf.Stanhill , 1978). This is considered to be aconfirmatio n ofth e realistic value ofth ecalculate d Nizzana yield. For it israthe r unlikely that any significant fertilization could have been accomplished inantiquit y on the runoff farming fields ofth e central Negev desert. Notwithstanding theuniqu e system of runoff farming,th edeser t environment does have a limited carrying capacity with regard toth eamoun t ofpeopl e and

91 accompanying animalstha t can besupporte d per unit ofdeser t area. Moreover, extensive livestock rearing, common forari d regions, isno t quitea syste m which allowsmanur e tob econcentrate d forus eo n thearabl e lands for food crop production. Therefore, apart from what theanimal s drop during their allowed grazing ofcertai n stubble fields,n oanima lmanur ewa sprobabl y added toth efields . It isclea r from thestudie sb y DeWi t (1958)an d hiscolleage s (VanKeule n and DeWit , 1975; VanKeulen , 1979, Penning deVrie san d Djitaye, 1982; Penning deVries , 1983)tha t lack of fertilization is, apart fromwater ,th e main reason oflo wyield s inari d regions. Thisi sals o theconclusio n from Stanhill (1978)i nhi sresearc h about anautarki c agro- ofa typical fellah's farm inth eYizree lValle y of Israel, based ondat a from the first decades of the20t h century. The average rainfall inth eare a isa t present about 483m m peryear . Averagewhea t yields amounted to53 1kg/h a (Table14 , last figure), whereby wheat was themos t important crop in this agro- ecosystem. Stanhill (1978) concludes "that themai n factor limiting wheat yieldso nth e fellah's farmwa s thecomplet eabsenc eo f fertilization".

WHEAT YIELDSFRO M THEAVDA TEXPERIMENTA L RUNOFFFAR M

Thecomparativel y fewdat a acquired by Evenari andhi scolleage so nwhea t yieldsa t theAvda t experimental runoff farm areno t representative ofancien t runoff farming conditions, since modern fertilizerswer e used in all cases, except for the firstyear . Itseem snevertheles sworthwil e to quote their interestingresults :

1961/1962 SEASON (Rainfall 64.7mm ) The first floodwa so nDecembe r 6, 1961, and penetrated to adept h of 120 cm,whic h corresponds to200 0cubi c meter ofavailabl emoistur e per hectare or 200 mm ofeffectiv e rainfall. Thisprove d tob eth eonl y flood of the rainy season and thewhea tdevelope d without any additionalwater . Thewhea tvarie ­ tiesEthi t andFlorenc eAuror e 8/39 were sowno n 17-12-1961 by broadcasting on a field that had received an amount of 35cubi c meter ofmanur e per hectare. The plants emerged on29-12-1961 . Full ripening and harvesting of Florence occurred on 21-4-1962an d 20-5-1962, respectively; and ofEthi t on 30-4-1962 and 17-5-1962. The yieldso f the twovarietie swer e95 0kg/h a and 600 kg/ha, respectively (Evenarie t al., 1963).

92 1962/1963SEASO N (Rainfall29. 5mm ) The research programfo rfiel dcrop swa sno tcarrie dou tbecaus e of the drought conditions. Allavailabl ewate rwa sdiverte dt oth eorchar dan d the range plantsan dth eplot sfo rth eannua lfiel dcrop swer elef tdr y (Evenari etal. , 1964).

1963/1964SEASO N (Rainfall183. 3mm ) Therewer ethre elarg ean dprolonge d floodsi nDecembe r1963 ,o n1-4 ,10-1 1 and30-3 1o ftha tmonth .Durin geac ho fthes eflood sth efiel dcro pplot swer e inundated for24-4 8hours . Theplot sreceive d920 0cubi cmete ro fwate r per hectare, whichi sequa lt o92 0m mo feffectiv erainfall . Moisturefille dth e soilsevera ltime st oa dept ho fmor etha n3 meter .Prio rt oseedin gth eplot s were preparedwit ha roto-tille r5 day safte rth e(second? )floo dan d ferti­ lized with30 0kg/h aammoniu msulfat e(21 %N )an d80 0 kg/hectare superphos­ phate (15%P 205). Twowhea tvarieties , FlorenceAuror e8/3 9 andEthit , were sown on25-12-196 3i nrow s2 0c mapart , witha see drat eo f13 0kg/ha . Good but slowemergenc estarte d1 1day safte rsowing ; birddamag ewa ssever e and couldno tb econtrolled .Th edate so fful lripenin gan dharvestin go fFlorenc e are6-4-196 4an d20-5-1964 ,respectively ;an do fEthi t14-4-196 4an d8-5-1964 . Theyiel do fFlorenc ewa s195 0kg/h aan do fEthi t139 0kg/h a(Evenar ie t al., 1965).

1964-1965SEASO N (Rainfall144. 2mm ) Sinceth efirs tfloo do fth erain yseaso nwa slate , on12-1-1965 ,n owhea t was sown thisyear . Runoffflow swer erecorde do n1 4day s(Evenar ie t al., 1968).

1965-1966SEASO N (Rainfall91. 5mm ) An early floodoccurre do n5-10-1965 . Thefiel dplot sfo rwhea t received prior to sowing 500kg/h asuperphosphat e (26%P 205 ), 400 kg/ha ammonium sulphate(21 %N) , and4 0cubic-metre/h aorgani cmanure .Sowin g (150kg/ha )o f theFlorenc ewhea tvariet yan dth eTheser a4 1dwar fvariet ywa scarrie dou to n 23-10-1965,b ywhic hth elatte rvariet yreceive da nadditiona lto pdressin go f 200kg/h aammoniu msulphat e(21 %N) . Harvestingtoo kplac eo n 7-5-1966, but the publishedresult sar esomewha tconfusing : p. 27mention sa yiel do fth e Florencevariet yo f438 0kg/ha , whereastabl e8 5give sa figur eo f267 0kg/h a forFlorenc ean d438 0kg/h ao fth eNanasi twhea tvariety , notmentione d ear­ lier.Th eyiel do fTheser a4 1i sno tstate d (Evenarie tal. ,1968 )

93 1966-1967SEASO N (Rainfall80. 7mm ) Seven very small floodsoccurre d and field cropsdi d not develop at all during this season (Evenarie t al., 1968).

ESTIMATED WHEAT YIELDSAN D RELATED POPULATIONLEVEL SA THORVA T HALUQIM

In the only year that nochemica l fertilizerswer e given to the wheat sown at theAvda t experimental runoff farm, but "only"3 5cubic-metre/h a of manure, yieldswer e60 0kg/h a and95 0kg/ha , respectively,fo rtw o different wheat varieties. It isunlikel y that such an amount ofmanur e could have been supplied yearly toth eancien t seedlands inth eRunof fFarmin gDistric t of the Negev inantiquity . During the6 years from 1961/62t o1966/67 , nowhea t was sown or harvested during 3years , for lack of runoffwate r or the first seasonal flood being too late.Th e averagewhea t yield atth eAvda texperimen ­ tal runoff farm during these 6consecutiv e years, including the yearswithou t any yield,i s99 5kg/ha ,obtaine d with the stated amountso fchemica lfertili ­ zersan dmanure . Average runoff water supply to the runoff farming system in the Nizzana desert area (present annual rainfall about 86mm )durin g the 7th century A.D. might also havebee n200-50 0m m in the year that the yieldswer e recorded. It seems unlikely that fertilization wasmuc h better inthi sdeser t region than in the semi-arid YizreelValley . Hence itseem sals o unlikely that wheat yields weremuc hhighe rtha n50 0t o60 0kg/h a inth eRu r ffFarmin g District during the 7thcentur y A.D. The averageyiel d of64 6 kg/ a, calculated from theNizzan a papyri on the suppositions described above (Table-14), does not seem too low an estimate,therefore . If conditions during the functioning ofth e runoff farming village at Horvat Haluqim, roughly 3000year sago , weremor eo r less similar to the farming situations decribed above, the 2.07 hectareo fterrace d fields inth e three wadis may haveyielde d some 1242k go fwhea t per year, based on an assumed average yield of60 0kg/ha . This quantity hast ob e reduced by the seed stock required fornex t year (about 200kg )an d post-harvest losses (Hall,1970 ;Va n derLee ,1978 )o fsom e 10 %(12 5 kg), which leaves anestimate d 917kg/yea r forhuma n consumption. Broshi (1980) takes the figureo f25 0k g grain/person/year as an overall minimum average required forhuma n subsistance. Theseaspect s ofhuma n food requirements are treated inmor e detail inchapte r 11. Thus only about 4

94 people could have lived onth e food production ofth ethre eterrace d wadis. Even assuming thepossibilit y ofhig h averageyield so f 1200kg/ha , not more than 8peopl ecoul d havebee n nourished by theloca lproduce . Aspas tpopulatio n figuresa tHorva tHaluqi mwer ecertainl y higher, consi­ dering theamoun t ofhouse san d other buildings (Cohen, 1976), theconclusio n has to betha t eithermuc hmor eterrace d wadisi nth evicinit y of the site were incorporated inth e local food economy, or foodwa sregularl y supplied fromoutside ,e.g . fromth ewette r northernpar t ofth ecountry .

95 1. <% HAR --« ». HAMRAN

/ < / .

\ \ \\\ \ / • \ » w

IV

k \\\r K V-

HAR HORSHA

'tf'ft ESCARPMENT 0 1 2 3 4 5 O SPRING 1 I I I I I km • TELL FIGURE 11.Are ao fKadesh-Barne aan dNaha lMltna n A NAHAL MITNAN SITE

sV iv «—y* ;V.

• ! - \ r\r FIGURE 12.Are a 4*X&. of Nahal Mitnan '-^^„. *5Mf /'^ inmor edetail . ST*^ --1^ -fe 11 2 • FARMSTEAD J km *VUc TERRACED WADI ^/* 600 m Contour

96 7 Nahal Mitnan

ENVIRONMENT

The terraced wadio fNaha l Mitnan issituate d closet o the border with Sinai (Egypt), ata distanc eo fabou t 37k msouth-wes t ofHorva t Haluqim, asth ecro w .Som e9 k mfarthe r toth esouth-west ,alread y inSinai ,lie s the Valley ofEi ne lQudeirat , considered tob eth eare a of Kadesh-Barnea (Figure 11),extensivel y dealt with inth enex t chapter. Nahal Mitnan isa tribuatar y ofNaha l Horsha, lying inth eLoz-Horsh a zone ofth erunof f farming district, southo fth eNizzan a zone (Kedar, 1967).Th e wadi flows from south-west tonorth-eas t andi sabou t3. 5k mlong , over which length iti sjoine d byit sow ntributarie so fth efirs tt othir d order (Figure 12). Nahal Mitnan isterrace d over mosto fit slength . Theterrace d partso f the wadi descend from about54 0m t o47 5m , wheni tjoin sNaha l Horsha. Its average gradient isnearl y 2 %. Many ofit stributarie sar e also terraced. Nahal Mitnan drains part ofSheluha t Kadesh-Barnea, a mountainous range situated toth enorth-west ,whic h risest oa naltitud eo f68 2m . This range formspar to fa comple x folding structure, together with Har Hamran (704m )t oth eeas t (Figure 3), that seemst ob econnecte d with the east-west trending Saad-Nafha-Hallal fault lineament (Bartov,1974 ;Zilberman , 1981). A structural saddle, through which Nahal Mitnanan dNaha l Horshaar e able to flow northward, separates Sheluhat Kadesh-Barnea from Har Hamran (Zilberman, 1981). The area drained byNaha l Mitnan islargel y composed ofwell-bedde d lime­ stones, with some silicified horizons,belongin g toth eNaha l Yeter Formation ofMiddl e Eocene age(Benjamini , 1979; Zilberman, 1981). TheEarl y Eocene

97 Nizzana Formation isexpose d along the entire south-eastern hillslope flank of Sheluhat Kadesh-Barnea, built of and chalky limestone with some chalk lenses. The soft marlso f theLat e Cretaceous toEarl y Tertiary Taqiye Formation, cropping out just over the top of this mountainous range (Zilberman, 1981), are apparently not drained by upper, first order,tributa ­ ries of NahalMitnan ,wit h a few possible exceptions. The average annual rainfall forth e area isestimate d at about 75 mm (p. 110). The hillsides are characterized by shallow desert lithosols and bare rock outcrops, which support a very scattered shrub vegetation dominated by Zygophyllum dumosum, Reaumurianegevensis , and Artemisia herba-alba (Danin, 1983). Most vegetation isconcentrate d in the terraced wadis, dominated by the Thymelaea hirsuta -Achille a fragrantissima association. The alluvial loessial soils in two adjacent wadi terraces were investigated by the author, the results ofwhic h arepresente d below.

ARCHAEOLOGY

A farmstead, situated along Nahal Mitnan,a t adistanc e ofsom e 2k m upstream (south-west) from the confluence with Nahal Horsha, was excavated by Mordechai Haiman (1982). The farmstead is located on anatura l terrace south of the stream bed ofNaha lMitnan , at an altitude ofabou t 510m , near the edge of limestone hillsmad e up by theEocen e Nahal YeterFormation . As the farmstead lies somewhat higher than thewadi-be d below, its former inhabitants could keep an eye onth eterrace d fields (Photo17) . The following data from the excavated farmstead, which has a size of 50x 20m , are quoted fromHaima n (1982): The farmcomprise s 11room s and court­ yards. The rooms aresquar e (4x A m), whereas thecourtyard s are rectangular, measuring about 7.5 x5. 5 m.Th e walls, 60c m wide,wer e built of two parallel rows of limestone blocks (30x 2 0cm )wit h a filling in the intervenient spaces.Haima n noted thehig h building quality with carefully laid courses and precise rightangles . The pottery found at the site, e.g. delicate bowlspainte d with a net design,cookin g pots,juglets ,an d an undamaged oil lamp,date s to the 7th and 8th century A.D. AnArabi c inscription was found on both sides of a glass disc. Other finds included fragments of iron and tools,morta r pieces, pounders, grinding stones, and a large number of millstones, which might suggest that cereal food crops were an important part of local runoff agricul-

98 ture in the terraced wadi systems. It seems that aByzantin e to Early Arab age can be attributed to the farmstead, aswel l as toothe r complexes along the banks of Nahal Mitnan and its tributaries (Haiman, 1982).

*?. *-

PHOTO 17. Nahal Mitnan with the Byzantine-Early Arab farmstead in the fore­ front, overlooking the terraced fields toward the hills ofShluha t Kadesh- Barnea. Theda m and terraced fields nearest toth e farmstead were investigated (pointed out by the arrow).

STRATIGRAPHY AND SOIL DEVELOPMENT IN BYZANTINE WADI TERRACES OF NAHAL MITNAN

STRATIGRAPHY The terraced stream bed ofNaha l Mitnan is flanked on both sidesb y natural wadi terraces,whic h presumably datebac k to theLat ePleistocene . The fine loessial sedimentso fthes enatura l terraceswer edeposite d when the wadi bed was situated ata highe r level than today. Aseroze m soil with a well developed calcic horizon has formed in these natural terrace deposits. The farmstead is actually located on the upper part ofsuc h a Late Pleistocene terrace near theEocen e hillsideedge .

99 *&

b'SPo^Wi M-C V»c5& • • 66 IM-C • 11 n,E IV-C |—| 110

O O O ° ° © °

°oW GRAVEL LAYER 50 100 SAMPLE • =L_ ==) cm

FIGURE 13. Stratigraphy ofa check-da man dtw oadjacen t terraced fields in thewad ibe do fNaha lMitnan .

Two pitswer e excavated intoth e man-made historical terraces within the wadi bed,o nbot h sideso fa check-da m thati slocate d nearestt oth e farm­ stead described above. The dam, built forrunof f farming purposes,i si nal l likelihood relatedt oth e adjacent farmsteadan dseems , therefore, to date back toth eByzantin e period. Onlyon e small pieceo fribbe d pottery, appa­ rently Byzantine,wa sfoun d during the excavations insideth etw o terraces. The dam is siltedu punt oit s brim with respectt o the upper terrace, whilsti trise s some4 0c mabov eth esurfac eo fth e lower terrace, downstream ofth e dam.A sth eprim e objecto fth e excavation wast ostud yth esedimentar y stratigraphy andsoi ldevelopmen to fth eman-mad e terraces, theda mwa s not excavated from insidean dit s architectureno tthoroughl y investigated.I t is, however, quite clear that the building styleo fthi sda mi smuc h more precise and orderly than thedam sa tth esit eo fHorva tHaluqim , whichma ydat e back toth eIro nag eo reve n earlier. The pit,excavate d intoth eterrac e downstream ofth eda m (Figure13) , surprisingly revealeda hidde n lower parto fth e dam, buried below alluvial

100 loessial wadi sediments. The dam is thus more complex and wider than initial­ ly anticipated, having awidt h ofabou t 2meter . This lower part of the dam is 110c m high, built of5 course s of limestone blocks, arranged in neat layers (Figure 13).Th e base ofth e lower dam rests on alluvial loessial sediments.A gravel layer appears some 10c m below the base ofth edam . The pit was not excavated beyond this gravel layer. The upper part of the alluvial deposits exposed inthi s pit,includin g a thin gravel layer at a depth of 15-25cm , are obviously younger than the lower dam itself. The other pit, excavated into the terrace upstream from the dam (Figure 13), exposed the base of the upper dam at a comparatively shallow depth of about 40 cm below thepresen t surface ofthi s upper terrace. A thin gravel layer occurs at a similar depth of38-4 3 cm. As the terrace surface upstream from the dam isabou t 40c m higher than the terrace surface downstream from the dam, it thus seems that these upper gravel layers, found in both pits, constitute one and thesam e stratum. The sedimentary stratigraphy and appea­ rance ofth e dam complex suggest that the upper part of the dam was built ina later stage. The reason for thisadditio n was apparently the gradual silting up of the terraced field unto the upper courses of the lower (then only)dam . The upper gravel layer,indicativ e ofmor e rapid stream flow inth ewadi ,wa s apparently deposited at theen d ofthi s first aggradational period since the wadi was terraced. In this stage the holding up function ofth e dams, by which the runoff floods were arrested and sedimentation induced, had apparently been overcome. Hardly any runoff flood waters could have been retained on the field, as the dam wasnearl y buried below the aggradational deposits. This necessitated the building of an addition to the dam on top of the alluvial sediments that had accumulated. However, the previous process repeated itself (thepas t is the key to the future!) as alluvial sedimentation subsequently raised the level of the terraced field unto thebri m of the upper dam. This is the situation at present. It isnoteworth y that more alluvial loessial sediment has accumulated upstream from the dam on top ofth e upper gravel layer, than downstream from the dam (Figure13) .

SOIL DEVELOPMENT The sediments downstream of the dam are somewhat finer textured than up­ stream, presumably because ofhydrauli c differences inth e flow of runoff water on both sides ofth edam . The alluvial soils in the terraced fields of

101 Nahal Mitnan, which accumulated and developed duringan dsinc eth e Byzantine period, contain more clay thanth ealluvia l soils in the first order wadi of Horvat Haluqim, apparently terraceddu ­ ringth eIsraelit e period.

TABLE 15.Granulometr yan dsoi l chemistry ofalluvia l loessial sedimentsi na n ancient runoff-farming terracei nth ewad ibe do fNaha l Mitnan. 1)Terrace d field upstream froma check-da mo fapparen t Byzantine origin.

HORIZON DEPTH CLAY SILT SAND TEXTURE GRAVEL CARBONATES <0.00S 0.002- 0. D63- >2m m CaC03 mm 0.063 2 mm (cm) (X) (%) (X) (X) (X)

A 0-15 21.1 56.1 22 .8 SILTLOA M 1 43.0 C 25-38 19.5 47.8 32 .8 SILTLOA M 0 45.4 II-C 38-43* GRAVEL III-C 47-58 26.7 52.6 20 .7 SILTLOA M 2 44.9 III-C 62-70 27.2 51.8 21 .0 SILTYCLA Y LOAM 1 45.5

HORIZON DEPTH pH EC(1:1) SAR ORGANIC NITRO­ TOTAL CARBON GEN PHOSPHORUS (cm) (mS/cm) (X) c (X)N (x)P 2O5 A 0-15 7.4 0.45 5.8 0.79 0.06 0.16 C 25-38 7.4 0.34 3.9 0.56 n.d. 0.13 II-C 38-43* III-C 47-58 7.4 0.46 5.9 0.46 n.d. 0.15 III-C 62-70 7.4 0.54 7.0 0.58 n.d. 0.15

HORIZON DEPTH SOLUBLE SALTSI N 1: LSOIL/WATE REXTRAC T (meq/kgsoil ) 2 (cm) Ca2+ .i2 + ..+ K+ HCO3 CI" so ; NO3 Mg Na A 0-15 1.0 <0 .10 4 .2 <0.1 4.3 0.4 0.6 0 C 25-38 0.7 <0.0 1 2 .4 <0.1 2.7 0.5 0.6 <0.1 II-C 38-43* III-C 47-58 1.6 <0 .10 5 .3 <0.1 2.4 3.2 1.1 0 III-C 62-70 1.6 <0.1 0 6 .3 <0.1 2.6 3.2 1.8 •0

Percentages inweigh t% o fth efin e earth< 2mm ; only graveli nweigh t %o f the total soil. EC= electrica l conductivity. SAR= sodiu m adsorption ratio, n.d.= no tdetermined . *Grave l layera t38-4 3c mdepth .

102 TABLE 16.Granulometr yan dsoi l chemistry ofalluvia l loessial sediments ina n ancient runoff-farming terrace inth ewad ibe do fNaha lMitnan . 2)Terrace d field downstream from acheck-da mo fapparen t Byzantine origin.

HORIZON DEPTH CLAY SILT SAND TEXTURE GRAVEL CARBO­ <0.002 0.002 - 0.063- >2 mm NATES mm 0.06]( 2 mm CaCO-7 (cm) (%) (%) (%) (% (%) A 0-1 2 25.0 54.6 19 .6 SILT LOAM 2 40.9 II-C 15-2 5* GRAVEL III-C 30-4 0 25.9 55.2 18 .9 SILT LOAM 0 41.1 III-C 57-6 6 36.1 49.6 14 3 SILT\ CLAYLOA M 0 44.1 III-C 77-8 6 34.5 56.6 8 .9 SILTV 'CLA YLOA M 0 44.2 III-C 95-105 28.2 47.3 24 6 CLAY LOAM 4 43.9 IV-C 110-120 21.8 53.7 24 5 SILT- LOAM/GRAVEL 15 46.8

HORIZON DEPTH pH ECU :1) SAR ORGANIC TOTAL CARBON PHOSPHORUS (cm) (mS/cm) (%) C ;%)P 205 A 0-1 2 7.4 0.7 0 4.4 1.37 0.18 II-C 15-2 5* III-C 30-4 0 7.4 0.3 0 2.9 0.73 0.15 III-C 57- 66 7.4 0.2 8 2.9 0.72 0.18 III-C 77-8 6 7.5 0.6 3 6.2 0.67 0.18 III-C 95-105 7.4 0.2 8 4.1 0.67 0.14 IV-C 110-120 7.4 0.2 5 2.9 0.72 0.11

HORIZON DEPTH SOLUBLE 5ALTS IN1: ] LSOIL/WATE REXTRAC T (meq/kg soil) „ 2+ 2 + 2 (cm) Ca ylMg Na+ K+ HC0~ CI" so ~ NO: 4 3 A 0-1 2 2.8 <0 .10 5.3 0.2 7.5 0.7 0 .7 0 II-C 15-2 5* III-C 30-4 0 1.1 <0 .10 2.2 <0.1 3.3 <0.1 0 .2 0 III-C 57-6 6 1.1 <0 .10 2.2 <0.1 3.0 <0.1 0 2 0 III-C 77-8 6 2.0 <0 .10 6.3 <0.1 2.7 4.4 0 .9 0.3 III-C 95-105 0.8 <0 .10 2.6 <0.1 2.5 0.3 0 4 0 IV-C 110-120 0.8 <0 .10 1.8 <0.1 2.4 0.1 0 3 0

Percentages inweigh t %o fth efin e earth <2mm ; only gravel inweigh t% of the total soil. EC= electrica l conductivity. SAR= sodiu m adsorption ratio. * Gravel layera t15-2 5c mdepth .

The soilsi nNaha l Mitnanar evirtuall y freeo fsolubl e salts. Theelectrica l conductivity rangesi nbetwee n0.3 0an d0.7 0mS/cm . Sodium isth epredominan t cation throughout theprofiles , whilst hydrogen-carbonate isusuall y dominant amongst theanions , often exceeding theamoun to fchloride , particularlyi n

103 the top-soil. As HC03 isgenerall y the dominant anion in rainwater (Yaalon, 1963, 1964; Nativ et al., 1983)an d runoff floodwaters (Nativ et al., 1983), there appears tob e acausa l relationship in thisrespect . The terraced wadi soils atNaha lMitna n are considerably more calcareous, containing an average amount of4 4 %carbonates ,tha n the sampled agricultural wadi terrace at Horvat Haluqim,whic h contains 30% carbonates on theaverage . The difference is largely attributed to the different lithology in therespec ­ tivecatchments . Comparatively softer limestone rocks with more intervenient chalk layers make upth e catchment ofNaha lMitnan , whereas harder Turonian limestone rockspredominat e in the catchment area ofth eHorva t Haluqim wadis. The terraced wadi soils ofNaha l Mitnan have a low organic C and nitrogen content, typical forari d regions. The amountsconcerne d are slightly more advantageous, however, than in the investigated agricultural wadi terrace at Horvat Haluqim, perhaps related to a comparatively better soil moisture regime in the more recentpast . The amount of total phosphorus in the studied soilso fHorva t Haluqim is, however, considerably higher than in Nahal Mitnan.Pas t human activity on the former terrace appears to have been more pronounced, as indicated by the reported finds. The phosporus distribution in Nahal Mitnan is nevertheless internally consistent with the past influence ofman , as the content of horizon-IV, which predates human activity, iseve n lower than the overlying layerstha t developed after the beginning ofrunof f farming.

The stratigraphic picture of the investigated check-dam and adjacent terraced fields in thewad i bed ofNaha l Mitnan can be summarized as follows, in accordance with thehorizo n nomenclature used in the Tables 15an d 16:

I. The present surface layers of the terraced wadi (Aan d C horizons), consisting of alluvial sediments that accumulated after the construc­ tion of the upper part of the check-dam. II. The upper gravel layer, which was deposited before the upper part of the dam was added. III. Alluvial sediments that accumulated after the terracing of Nahal Mitnan for runoff farming purposes, i.e. after the building of the lower part ofth e check-dam across this stretch ofth e wadi bed. IV. Alluvial sediments predating the terracing ofth ewadi .

104 8 Kadesh-Barnea :environment ,histor yan dth etel l

ENVIRONMENT

GE0M0RPH0L0GY AND GEOLOGY The Highlandso fth eCentra l Negev continue forsom e 10-20k macros s the border into Egypt. Further toth ewes t stretchesth ebarre n expanse of Northern andCentra l Sinai, characterized bywid e plains, plateaus,valleys , individual hills, andisolate d mountainous domes. Viewed from Sinai, the Central Negev highlands, ranging inaltitud e from about40 0t o100 0 meter, look likea nimpressiv e mountainous plateau. The Kadesh-Barnea areai ssituate d atth ewester n edgeo fthes eHighlands , about6 t o1 0k minsid e Egypt (Figure2 , 14).Th esprin go fEi ne lQudeirat , arising inth egorge-lik e valley ofWad ie lQudeirat , isth emos t important feature ofth eKadesh-Barne a area interm so fman-lan d relationships. It is apparently the richest spring ofth eentir e Sinaian dNege v desert region, producing watero fa relativel y good quality. Theaverag e discharge of the spring is about6 1cubi c meterso fwate rpe rhour , flowing as a perennial stream inth echanne lo fWad ie lQudeira t overa distanc eo f2 km . The rather narrow valley plain, inwhic hth ewad iha scu ta steepl y banked stream channel, rangesi naltitud e from about 350m a tit smouth , near the confluence with WadiUr nHashim , toabou t40 5m a tth eactua l site of the spring's major outlet.Th evalle y isenclose d bya serie so fspine y ridgesan d escarpments that emenate fromth ehighlands . Gebele lUda h (522m )limit sth e area inth esouthwest , whereas Gebele lQudeira t (500m )bound sth e Kadesh- Barnea valley toth enort h (Figure 14). Thewidt ho fth evalle y plain ranges from about40 0m a tit swester n entrancet osom e4 0m nea rth espring .

105 HAR HAMRAN

<"*N .&>*

^ Jy ;,_t

•; ..... „„—, U » » \ V% > i ' J\ V--' .X^di \ \X \ M V? \ V\ \ \\ tn \_ r \

HAR HORSHA ^H ESCARPMENT 2 O SPRING km Q TELL A NAHAL MITNAN SITE

FIGURE 14. Thecatchmen to fNaha lKadesh-Barne aan dth e valley oasiso fEi n elQudelrat ,showin gth elocatio no fth esprin gan dth etell .

Viewedi na somewha t wider geological perspective, Gebele lQudeira t forms part of acomple x folding structure relatedt oth e Saad-Nafha-Hallal fault lineament. Thisi sth emain , east-west trending, linear fault systemi nth e northern parto fth e so-called Central Negevan dSina i shear belt,situate di n between the two main faulting zoneso fth eNegev-Sina i region: theDea d Sea Rift Valley in the eastan dth e Suez Rift in the west (Bartov,1974 ; Zilberman, 1981). Theridg eo fGebe le lQudeira t continues eastwarda sGebe l elEi nan dSheluha t Kadesh Barnea,whic h together form the southern flanko fa complex, east-west trending, fold structure that includesHa rHamra n farther toth e east.A structura l saddle,throug h which Nahal Horshaan dit s tributary

106 NahalMitna n runnorthward , separates Har Hamran from Sheluhat Kadesh-Barnea (Zilberman, 1981). Just north ofGebe l el Qudelrat and Gebel elEl n existsa complex fault system, which is related to the folding structure described above, forming part ofth eSaad-Nafha-Halla l lineament (Bartov, 1974).

PHOTO 18. The Kadesh-Barnea valley and the beginning ofagricultura l cultiva­ tion (date-palms)nea r the broken Jarvis Dam, some A00m downstrea m from the spring proper. Themountainou s ridge south of the valley isbuil t of Eocene calcareousrocks .

The steep, mountainous ridges north and south ofth eKadesh-Barne a valley are mainly built ofhar d limestones, aswel l as chalk, ofth e Eocene Avdat Group, overlying the soft marl and chalk ofth eMoun t Scopus Group in the subsurface. The Lower toMiddl e Eocene Nahal Yeter Formation (Benjamini, 1979) forms the upper part of the ridges, overlying the Lower Eocene Nizzana Formation,whic h in turn overlies the Paleocene TaqiyeFormation . The Nahal Yeter Formation iscompose d ofwell-bedde d limestone with some silicified horizons. Large-scale deformations characterize the top of this formation, formed by collapsed units ofwell-bedde d limestone layers,10-2 0m thick. These sedimentary structures may have been caused by Eocene earthquakes

107 (Zilberman, 1981). The NizzanaFormatio n consistso flimestone s and chalky limestonewit h somechal k lenses. The layersexhibi t smallt omoderat eintra - formational sedimentary disturbances,lik esmal l folds,detache d layers,sedi ­ mentary conglomerates and small sedimentary faults. The underlying Taqiye Formation ismainl y composed ofmarl , cropping out atth e lower parto f the northern and western slope ofGebe le lQudeirat , aswel l as across the entranceo f theKadesh-Barne a valley near theconfluenc e with WadiUr nHashim .

GEO-HYDROLOGY The main upper aquicludeo f theCentra lNege v Highlands is the Taqiye Formation, mainly consistingo fimperviou smarl . This formationunderlie s the entire Eocene Avdat Plateau fromKadesh-Barne a inth ewes t until the Zin Valley inth eeast . Runoff rainwaters thathav e infiltrated into the under­ ground may concentratea s shallow groundwater on top ofth eTaqiy e marls.A sa result ofthi shydro-geologica l situation, some springsaris e at favourable locations on the outskirtso fth eHighlands , nearoutcrop so f the Taqiye Formation. Thegorge-lik e valley ofWad ie l Qudeirat isincise d into theEocen e Avdat Group and actually cuts into themarl so fth eTaqiy eFormatio n some 3 km downstream fromth epoin t where the spring ofEi ne lQudeira t emerges in the wadi bed. The relation seemsobvious ,bu t the local foldan d fault structures of theSaad-Nafha-Halla l lineament,t owhic h theKadesh-Barne a area isconnec ­ ted,ma y alsohav e an important influenceo nth epositio no fth espring . Thelatte r view seemsconcurren t withth e existenceo fth e nearby spring of Quseima, 7k m west-northwest ofEi n elQudeirat , situated on theSaad-Nafha - Hallal lineament. This spring appears tob ealmos t ascopiou s as Ein el Qudeirat, having adischarg e ofabou t 50-60cubi cmete ro fwate r per hour,a s reported byKarscho n (1984). Thewate r ofth eEi nQuseim a spring appears tob e of even better quality thanth e spring ofEi n el Qudeirat. Its reported chloride content of250-30 0mg/lite r (7.1- 8. 5 meq/liter)i sonl y half the quantity occurring inth e latter spring. Data about thewate r ofEi n el Qudeirat areshow n inth e following table. Thesprin g is flowing allyea r round and thedifference s indischarg e between the variousmonth san d yearsar e relatively small, ranging in between4 8an d 73 cubicmeter so fwate r per hour. Theaverag e discharge isabou t 61 cubic meter/hour. This suggests aconsiderabl e relaxation time between rainfall events in thecatchmen t area and spring flow,probabl y caused by thebufferin g effect ofa sizeablenatura l storage of shallowgroundwater .

108 TABLE 17. Discharge dataan dchemica l analysiso fth e spring water ofEi ne lQudeirat . (based upon data fromth eIsrae l Hydrological Service)

SAMPLEDAT E DISCHARGE (cubicmeter/2 4hours ) (cubicmeter/hour )

1926 1500 63 23-08-1951 1400 58 21-10-1969 1750 73 19-01-1970 1150 48 18-05-1970 1600 67 16-07-1970 1450 60 21-09-1970 1400 58 17-02-1971 1400 58

Ca Mg Na+ K+ HC0~ CI" SO^"

89 45 330 4 (mg/liter) 212 509 226 4.44 3.70 14.35 0.10 (meq/liter) 3.47 14.36 4.71

AGRICULTURAL EVALUATIONO FTH E SPRING WATER The chemical composition ofth e water appearedt ob e fairly constant during theyear s from 1926t o1971 . The chloride content ranged in between 500-580 mg/liter (14.1- 16. 4 meq/1). The amounto fdissolve d saltsi sabou t 1.42 g/1an dth eelectrica l conductivity (EC)o fth e spring wateri sestimate d at2 mS/cm . The sodium adsorption ratio (SAR)i s7.11 , showingth e sodicity (alkali) hazard ofth ewate rt ob elow . Thisi sconfirme db y the negative value (-4.67)o fth eRS C (residual sodium carbonatean dbicarbonate) , another criteriont ojudg eth e sodium hazardo firrigatio n water (ILACO, 1981). For irrigation purposesth e water from the spring can be classified as havinga medium-hig h salinity hazard (Thornean d Peterson, 1964). The wateri s suitablet oirrigat e cropswit ha moderat et ogoo d salt tolerance. Irrigation is allowedo nsoil s witha moderat et opoo r permeability.Th e crusting loamt o silt loam soilsi nth eKadesh-Barne avalle y are, therefore, suitedt o this typeo firrigatio n water. Leaching with excess wateri srequired , however,t o preventa built-u po fsalt si nth eroo t zone.

RAINFALL The average annual rainfall todayi nth eKadesh-Barne a area has tob e estimated from datao fnearb y locations. Mean annual rainfall at Quseima, situated about7 k mwes to fEi ne lQudeira ta ta naltitud eo fsom e 300m , is

109 64 mm (Israel Meteorological Service, 1967). Nizzana receives a mean annual rainfall of8 6m m and lies 26 km north ofEi n elQudeira t at an altitude of 250m . The average rainfall in theKadesh-Barne a area (altitude of the Ein el Qudeirat valley ando fth e surrounding ridges and highlands is 350-425 m to over 500m , respectively) can therefore be estimated at about 75mm/year .

SOILS and VEGETATION In terms ofman-lan d relationships, theKadesh-Barne a valley isno t only endowed with acopiou s spring, flowing all year round, but also with culti­ vable soils formed in alluvial loessial sediments. These young alluvial soils show little or no profile differentiation in the field. However, on close examination, including microscope studies and chemical analysis, remarkable features of soil development were discovered in the valley soils adjacent to the steep banks of thenarro w wadi channel. Loessial serozems, usually exhibiting calcic, gypsic and saline horizons, are found on moreelevate d Pleistocene terraces in the valley. Calcareous desert lithosols and Brown lithosols occupy the surrounding hillslopes and highlands, covered by a desert vegetation dominated by Zygophyllum dumosum, Artemisia herba-alba,Salsol a tetranda,Suaed a vera var.desertis . The vegetation along the stream channel, fed by the flow of spring-water, includes Tamarix nilotica, Nitraria retusa, Typha australis and Phragmites australis (Danin, 1983,persona l communication, cf.Goldberg , 1984).

HISTORY AND GEOGRAPHY

Kadesh-Barnea appears to be the principal place where thepeopl e of Israel sojourned inth edeser t for some4 0 years under the leadership of (Num. 14:33-34), after from Egypt and before they entered the promised land. Most references in theBibl e toKadesh-Barne a are indeed rela­ ted to this period within theLat e Bronze Age (1550-1200 B.C.), when the Israelites "abode inKades h many days" (Deut. 1:46). However, the site is already mentioned in the time of Abraham, in theMiddl e Bronze Age, around 1900B.C. , when Kadesh isequate d with En-mishpat, "The Spring of Judgement" (Gen. 14:7). The location ofKadesh-Barne a can be inferred from thedescriptio n of the southern border of theLan d of Israel (Num. 34:1-6; Ezech.47:19 ; 48:28)an d of the southern border ofth e territory allocated to thetrib e of Judah (Josh.

110 15:1-4). Numerous scholars of past generations have attempted to identify biblical Kadesh-Barneawit h sites inSinai , theNegev , oreve n more distant places, as summarized by Cohen (1981). A general scholarly consensus emerged early in the 20th century, relating Kadesh-Barnea to theoasi s of Ein el Qude.irat . Woolley and Lawrence (1914-15)regarde d the region around Quseima (Fig. 2) with its springs and arable soil, at the crossroads oftw o of antiquity's major desert routes, as the likely area ofKadesh-Barne a from a more broad perspective. The sheltered valley-oasis ofEi n el Qudeirat, in the eastern part of the Quseima region,i sconsidere d by them asth emos t likely candidate for Kadesh-Barnea ina mor e restricted sense, by which they focussed special attention to the Tell. The authors made their observations in January and Febuary of 1914. The following quotations eloquently describe the area and someo fth e reasons for relating it toKadesh-Barne a (Woolley and Lawrence, 1936:88,87) :"Thes e roads running out to north, south,east ,an d west...togethe r with itsabundanc e of water and wide stretch of tolerable soil, distinguish theKossaim aplai n from any other district inth eSouther n Desert, and may well mark it out as the headquarters ofth e Israelites during their forty yearso fdiscipline... only in the Kossaima district are to be found enough water and green stuffs to maintain so large atrib e for so long, and that therefore the Wilderness of Zin and Kadesh-Barnea must be thecountr y of Ain el Guderat, Kossaima, Muweilleh, and AinKadeis" . In their description of the springs in theQuseim a region, they contrast the barren and rugged area around Ein Qadeis with themor e cultivable Quseima plain and the oasis ofEi n elQudeirat . The latter site is clearly distin­ guished asth e most outstanding spot in the entire region (p.75) : "In place of the solitary Ain Kadeis are AinMuweilleh , in a soft wady bed; Ain Kossaima, aplentifu l running ofwate r in the sand; and Ain Guderat, a great spring, not set in a dung-heap like Ain Kadeis,o r sand-choked like all other Negeb springs, but bursting straight from the rock, and running down a deep green valley of lush grass in swift irrigation channels, or in a long tree-shaded succession of quiet pools many feet deep. This plain about Kossaima runs from Muweillah on thewes t to the foot ofth e great pass of Ras Seram on the east. In fortunate years itmigh t bever y fruitful; and in the worst seasons itscrop s cannot entirely fail, thanks to the irrigated valley ofAi n elGudera t - theonl y stretch ofcorn-lan d under running water which we saw in all the southern waste."

Ill ARCHAEOLOGY OF THE TELL

Woolley andLawrenc e were the first scholars with solid archaeologi­ cal training to study the Tell in the valley of Ein elQudeirat , situated some 1600mete r downstream from the spring, "commanding the finest water-supply in all the desert" (p. 84). They described the tell asa little mound rising above the surrounding corn-fields. After brief excavations, they found the tell to be made up by the remains of an ancient fortress and contemplated the possibility that itmigh t have existed already during the time of theExodus , and perhaps be linked specifically to Kadesh-Barnea. The excavations by Dothan in 1956 further clarified the ground plan of the fortress,datin g itsconstructio n to the Judean kings ofth e 8th-7th centuries B.C. (Dothan, 1965). Thus, the tempting possibility posed by Woolley and Lawrence (1936:88) "that the fort was already built when Moses came" turned out to be unreal for this particular stronghold. A reliable stratigraphy was, however,stil l lacking for the tell. Rudolph Cohen conducted thorough excavations at the site during ten seasons from 1976-1982. These extensive investigations revealed the remains of two other fortresses. Thus the tell isno t made up by just one stronghold of one period, i.e. theLat e Iron age,bu t by three superimposed fortresses,eac h one built over the remains of its predecessor. A concise description of these fortresses, quoted fromCohe n (1981, 1983), followsa t page 116. The Early Fortress and anumbe r of related buildings and silos appear to be the oldest archaeological structures found at the site, attributed by Cohen to the 10th century B.C. (Cohen,1980 ,1981 , 1983).

CARBON-14 DATE (GrN-12330)- LATES T BRONZE TOEARL Y IRON AGE A charcoal sample was taken by the author in collaboration with Cohen from one of the lowermost ash layers, some 5mete r below the surface of the tell-mound. This layer seems stratigraphically connected with the destruction of the Early Fortress. The sample was investigated at the Isotope Physics Laboratory ofProf .Moo k (University ofGroningen ) and yielded a date of2,93 0 + 30B P inconventiona l Carbon-14 years (GrN-12330). When this date is trans­ lated into historical years, calibrated against the high precision curve of Pearson (1986), for reasons explained in the following section,th e resulting age of the ash layer is surprisingly early: 1255-1055 CalB.C . The natural level ofcarbon-1 4 in the atmosphere hasno t been constant with time, as first reported by De Vries (1958)wh omad e this important observation

112 PHOTO 19. The excavated tell in the valley ofEi n el Qudeirat, exhibiting primarily the outline of theUppe r Fortress (Late Iron Age). The stream channel carries thewate r from the spring along the southern part ofth e tell. indirectly through C-14 measurements of tree-rings from European and North American wood. A radiocarbon date, asa consequence , although consistent in itsow n timescale, cannot simply beplace d intoa historica l time framework. First the carbon-14date , expressed in conventional C-14 years, hast o be translated intoastronomica l years. The resulting calibrated C-14 date can then be correlated with historically dated events (Mook, 1983). In order tomak e this calibration possible, very detailed and precise C-14 analyses were carried out onwoo d of known astronomical age, dated bydendro ­ chronology. Research on oak wood from Ireland has resulted in ahig h precision radiocarbon calibration curve,publishe d by Pearson (1986). A conventional C-14 date isalway s presented with arang e of uncertainty, called standard deviation (sigma cr). The probability that the actual date is situated within this time period is 68% . Ifth e age range is enlarged to about 3sigma , the probability becomes virtually 100 %. It should also be realized that acharcoa l sample, in most cases, does not represent a single

113 year, likea tree-ring ,bu tcomprise sa certai n time-width.I na nidea l situa­ tion, theamoun to fdetai li na calibratio n curve oughtt ob eadapte dt oth e time-widtho fth eC-1 4 sample (Mook, 1983). when a radiocarbon datei stranslate d from conventional C-14 years into historical years, notjus tth emedia n yearo fth eC-1 4 date, butth eentir e sigmarelate dag erang eo fth e dateha st ob eplotte d againstth e calibration curve. Apart from sigma (cr),th estandar d deviationo fth ecalibratio n curve, being1 5years ,ha st ob etake n into accounta swel li norde rt odetermin eth e correctag erang e forcalibration : (a)Calculat eVo -2+ 15 2; (b)Th eresultin g number Xi sadde dt oth emedia n yearo fth e C-14 date, which givesth euppe r time limito fth e date's ranget ob ecalibrated ; (c)X i ssubtracte d fromth e median yearo fth eC-1 4 date, which givesth elowe r time limito f theC-1 4 datet ob ecalibrate d (Mook,persona l communication, 1985). Theag erang eo fth echarcoa l sample under discussion (GrN-12330;293 0+ 3 0 BP), tob euse di nth ecalibratio n plotting, is2965-289 5B Pi nconventiona l C-14 years. This sigma related time rangeo fth e datei splotte d (Fig.15 ) against thehig h precision radiocarbon timescalecalibratio n curveo f Pearson (1986). The resulting calibrated datefo rth e blackas hlaye ri n historical yearsi s1255-105 5 CalB.C .

Q_ m < 3100 >UJ-

i 3000

2965

2930 2895 j. \/"\ I | ^\ I | " I I I . 2800- I

1400 1300 1255 1200 1100105 5 HISTORICAL YEARS B.C.

FIGURE 15. Calibrationo fth e above C-14 date (GrN-12330) into historical years (apparently relatedt oth etim eo fdestructio no fth eEarl y Fortress).

114 Taken at face value this surprising outcomemigh t indicate that the Early Fortress isperhap solde r than anticipated. Ifth e ash layer is indeed related to the time of destruction of the latter stronghold, then the radiocarbon date of this ash layer suggests this to have occurred during the very end of the Late Bronze Age or Early Iron Age. The logical implication is, that the time of construction of theEarl y Fortress must beeve n older than the above C-14 date, pointing to theLat e Bronze Age,perhaps . The result of this radiocarbon date and its ramifications, touching upon the dramatic transition from theCanaanit e to the Israelite period in the Negev and Sinai, aswel l asth e outcome and stratigraphic context ofothe r C- 14 dates from the tell, will be discussed and evaluated in a joint publication.

The dated ash layer covered a soil profile with the following stratigraphy and characteristics:

A 0-10 cm Black ash layer. 10 C 10-20 cm Loessial soil,loam y texture,containin g 20 1-2 % ofwhit enodules . II-A(C) 20-30 cm Loessial soil,sand y loam texture,con ­ 30 oOo "o"° taining a fewsmal l pieces ofcharcoal . 0 o oo o o ° II-C 30-50+ cm Fine gravel mixed with sandy loam. o o 00 °0 ' O°„0 OB 50 *oo° °o o° The ash layer can be regarded asa disturbe d Ahorizon , overlying a C horizon that consists of loess with a loamy texture. About 1-2 % of white nodules, composed of gypsum or secondary carbonates,appea r in this Chorizon . These nodules were formed by natural soil forming processes under the influ­ ence of rainfall, which caused influx,dissolution ,concentratio n andrecrys - tallization of thesechemica l compounds in the upper part ofth esoil . It isobviou s that nodule formation could only takeplac ewhe n the soil was still situated at the former landscape surface, during the period before and/or in the time after the formation of theC-1 4 dated ash layer. With the building of the Middle Fortress, ifno t already during the time of the Early Fortress, the soil became buried below the rainfall penetration depth, which isabou t 30c m for aloam y soil in the average arid climate of the region. The next II-A(C)soi lhorizo n has atextur e of sandy loam. Itcontain s some small pieces of charcoal,probabl y older than the dated black ash layer. It is

115 apparently another buried Ahorizon . The charcoal remainsma y haveoriginate d from human activity atth esite , perhapsdurin g theBronz eAge , by which no artifacts were apparently left behind. Theunderlyin g II-Chorizo nconsist so fvirgi n soil, composed ofa natura l mixture of finegrave lan d sandy loam.

EARLYFORTRES S TheEarl y Fortressha sa n oval groundplan with adiamete r of 27m ,an d is similar to the other Israelite fortresses inth eCentra lNege v such as at HorvatHaluqim :p .69 ,phot o 12 (Cohen,1976 ,1980) . Theash-covere d floorso f thecasemat e roomscontaine d aric h assemblage ofpottery , attributed to the 10th century B.C. (Cohen,1981 ,1983) . Unlikeothe r similar strongholds inth e Central Negev, whichwer eno t rebuilt after theirdestruction , thesit e at Kadesh-Barnea isuniqu e inth e sense that ane w fortresswa serecte d over the ruinso fth e formeron e (Cohen, 1980).

MIDDLE FORTRESS TheMiddl eFortres sha d acompletel y different, rectangular,groun d plan with asiz e ofabou t6 0x 4 0m . Itha d a solid wall,abou t4 mete r thick,an d eight projecting towers. The fortresswa sentirel y surrounded by an earthen glacistha t rested against abuttres s wall, 2.50m high. Thiswal lwa s sur­ rounded by amoat , about 4m wide,excep t on the southern sidewher e thewad i channel functioned asa natura lmoat . Abi gcistern ,ca .1 0m acrossan d with a capacity for about 180cubi cmeter so fwater , wasuncovere d inside the fortress. Aplastere d conduit, running underneath thesouther n rampartwall s toward the sideo fth ewadi , facilitated the filling ofth ecister n with water from thespring .Silo san d granarieswer e also discovered (Cohen,1981 ,1983) .

UPPER FORTRESS The UpperFortres sa tKadesh-Barne a waserecte d ashor t time after the destruction of theMiddl eFortres san d its ground plan, already outlined by Woolley andLawrenc e (1914-15)an d Dothan (1965), isalmos t identical totha t of the latter.Casemat ewalls ,however ,replace d thethick ,massiv e walls.Th e earthen glacis wasadapte d toth ene w walls, whereby thebuttres swal l and moat continued to function, aswel l as thecister n and itsconduit ,tha t runs underneath thesouther nrampar twall s toward the sideo fth ewad i (Photo 20). On the ash-covered floorso f theth ecasemat e and interior rooms a large assemblage of ceramic vesselswa s uncovered, thewheel-mad e pottery being

116 ^Cti-V i

PHOTO 20.A plastered conduit,runnin g underneath thesouther n rampart wall of the Upper (and Middle)Fortres s toward the sideo fth ewadi , facilitated the filling ofth e largecister n inside the latter fortresses. characteristic for the 7tht o6t h century B.C.On e jarwa s still full ofburn t wheat. The third and last fortress atKadesh-Barne awa sprobabl y destroyed by the Babylonian army ofNebuchadnezza r at about the sametim e as the First Temple inJerusale m -58 6B.C . (Cohen,1981 , 1983).

PERSIAN PERIOD After the destruction of the UpperFortress , therewa sevidentl y limited reoccupation on the sitedurin g the Persian Period (587-332B.C. ) inth e form ofa nunwalle d settlement (Dothan,1965 ;Cohen ,1981 , 1983).

ENVIRONMENT ANDSITIN G OF THE TELL

The tell-mound rises some 6mete r above the surrounding valley plain, which is 164mete r wide at thispoint . Themountai n edge ofGebe le l Qudeirat boundsth e valley some 78m toth e north ofth etel l (Photo 19,22) ,

117 measured from the northern wall of the upper and middle fortress. The peren­ nial desert stream, fed by the spring, flows along the southern side of the tell (Photo 19). The mountain edge that bounds the valley in the south (Photo 18, 22) lies about 46m from the southern wall ofth euppe r and middle for­ tress. These fortresses themselves had awidt h of some 40m (Cohen, 1983), as a consequence ofwhic h the upper part of the tell isabou t 40m wid e as well. The tell is made upb y the remnants of the three fortresses that were successively built on the site. Probably throughout itshistor y the tell functioned as a sedimention trap for dust, mixed inwit h the remnants and debris of successive cycles ofbuildin g and destruction. Studies about soil development at the tell in relation to its stratigraphy will be published elsewhere. The tell is situated on a slightly elevated hillock, a stony remnant ofa Pleistocene terrace inth e middle of the valley of Ein el Qudeirat. This hillock isprimaril y composed ofcoars e gravel and well-rounded boulders up to 40c m in diameter (Photo 21), being somewhat cemented. These coarse sediments were deposited during theLat e Pleistocene, when the stream bed of the wadi was situated well above the present level ofth e valley.

PHOTO 21. Late Pleistocene deposits, consisting of slightly cemented coarse gravel and well-rounded boulders up to 40c m in diameter, below thebas e of the southern wall ofth e Upper Fortress.

118 '.fV'»-4>. "«St "

PHOTO 22. The valley ofEi n elQudeira t widens here,jus t downstream from the Jarvis Dam; the excavated Tell isvisibl e farther downstream, located in the centre of the valley oasis,mor e westward.

Goldberg (1984)foun d Middle Paleolithic discoid cores in a similar terrace remnant, presumably dating back 90.000-40.000B P (Schwarcze t al., 1980), about 1700m downstream from the tell, at the entrance of the valley near the confluence with WadiUr nHashim . Before the construction of the Early Fortress, the site must have looked like a low, natural hillock, rising some 3mete r above thegenera l level of the valley plain. This stony hillock constituted thenatura l foundation on which theEarl y Fortress and subsequent fortresseswer e built, right in the middle of the valley, yet firmly established on the stony underground (Photo 21)t owithstan d the seasonal floods.A shallow soil with a loam to sandy loam texture, found below some of the ancient man-made structures,covere d part of the stony hillock. It isnoteworth y that the ancients did not built their strongholds next to the actual spring itself, but farther downstream where the valley is wider (164m )an d suited for irrigation agriculture.Th e site selected by the people

119 who built theEarl y Fortressan dsubsequen t fortresses was apparently the only elevated stony foundation withinth eloam y valley plain, situateda ta very strategic position: about half-way between the entranceo fth e Kadesh-Barnea valley (which could bekep ti nvie w some 1500m westward ) and the actual location of the spring (some 1600m eastward). Thepreciou s water from the springi sflowin g naturallyt oth esit eo fth etel l all year round,carrie db y the narrow stream channel, whichi sincise d below thevalle y plainan dpasse s the tell just 10-20mete r southo fth emai n southern fortress wall. Afe w hundred meter farther downstream, however,thi s unique perennial streami nth e desert disappears below the surfaceo fth e wadibed . The positiono fth e tell thus hasth efollowin g advantages: 1. It commands thewesternmos t accesst oth e spring-water. 2. Iti ssituate d amidst suitable farming landsi nth ewide r parto fth e valley. 3. The elevated stony underground providesa fir m foundation against seasonal runoff floods. 4. Itgive sa vie w unto thewester n entranceo fth e valley.

*^A.D .7t hcentur yDam &Aquaduc t 0.5 1.5 2 km © Spring 0 AquaduCtS (1725-1425 Cal BC , 615-665 Cal AD) III Jarvis Dam —-* Wadi • Tell *•"•-—- Perennial Stream ^.Terraced Wadi for runoff farming

FIGURE 16.Th e valleyo fEi ne lQudeira to rKadesh-Barne a

120 9 Kadesh-Barnea: irrigation agriculture

SYSTEMSO FIRRIGATIO N AGRICULTURE INTH EKADESH-BARNE A VALLEY

The occurrenceo fa copiou s spring, whose wateri sflowin g freely all year roundi nth emiddl eo fth e desert, ina valle y also endowed with cultivable soils, maysugges tth epractis eo fagricultur et ob ea natura lan d simple matter. Inrealit yth e situationi ssomewha t morecomplicated , essen­ tiallydu et oobstacle so fa geomorphi c nature. Themai n problemsfo rfarmer s to practise agriculture inth evalle yo fEi ne l Qudeirat are essentially twofold: 1. The water from the spring arisesan dflow si nth e stream channel of Wadie lQudeira t (NahalKadesh-Barnea) , situated below the level of the cultivable valley plain. 2. Although theaverag e annual rainfalli nth e areai slow , being about 75m mtoday ,powerfu l runoff floodsma yoccasionall y developi nNaha l Kadesh-Barnea duringth e rainy season, causing damageo rdestructio n to irrigation systemsan d crops.

In ordert osolv e the first problem, thefarmer sha dt obrin g the water from the spring unto the levelo fth ecultivabl e valley plain. Theremain so f former and present irrigation systemsi nth e valley show two principal solutions: (a)AQUADUCT STHA T BEGINA TTH E NATURAL LEVELO FTH E STREAM BED, either having their intakeo fwate r nearth e main outleto fth e spring or farther downth e wadi bed. This typeo faquaduc t must initially havea gradien t which is less thanth eslop eo fth e stream channelan dvalle y plain.A sth eaquaduc t

121 descends less steep than the latter, itwil l seemingly rise with respect to the channel bluffan d will appear at some point downstream onto the level of the valley plain asa result. The point ofemergenc e ofth e aquaduct onto the valley plain isobviousl y the highest level where irrigation and agriculture can take place. From thispoin t the aquaduct may have the same, steeper, gradient as the valley. A low threshold-dam that raises the water-level somewhat behind itwil l considerably facilitate the intake ofwate r into the aquaduct(s), thus improving this system.

(b) THE BUILDING OF AMAJO R DAM that raises the water level until the height of the valley plain, asa significant artificial pond will form behind such an obstruction wall in the stream channel. Aquaducts can thusbegi n at the level ofth e valley plain right from their beginning, which may allow the valley plain tob e irrigated farther upstream, depending of course upon the location of the dam. This system enables a better control and more efficient use of the spring-water for agriculturalpurposes .

It seems obvious that aquaducts which have their beginning in thewad i bed (type a)ar e susceptible to destruction by powerfool runoff floods, occasio­ nally roaring through the stream channel in the rainy season. Aquaducts connected with high dams that raise the water until the level of the valley plain, can be built right from their beginning point ofwate r intake on the edge of the valley against the mountain side. Inthes ehig h dam systems the aquaducts are rather safe, but now the dam itselfma y be destroyed by runoff floods, historic examples of which willb e discussed in the following sections. At present the stream channel only occupies a small part of the Kadesh- Barnea valley. The channel isabou t 5t o 15mete r wide.I tha s steep banks and its bed iswel l incised toa depth of 3 to 6mete r below thepartl y cultivated valley plain, within historical times the stream channel has experienced dramatic changes,extensivel y dealt with in Chapter10 . Each irrigation system in the Kadesh-Barnea valley necessarily comprises two different aspects. The first aspect involves theconductio n or raising of water from the spring onto the valley plain, and has already been discussed. The second aspect involves the equal distribution of irrigation water over the agricultural fields. Possible distribution systemstha t wereprobabl y used at various times in thepas t can becharacterize d aswil d flooding, border, and basin irrigation (Booher, 1974).

122 These systems may have been integrated orcombine d in various localadaptations . The main irrigation aquaducts or ditches were ususually placed along both high edges of the valley plain, against the hill­ sides. By making outlets at desired points into these main aquaducts, the water would flow onto that part ofth e valley plain requiring irrigation.Eac h part of the valley plain is sloping down both laterally towards the stream channel and longitudinally in a general downstream direction. The farmer had several options in distributing the flow ofwate r as PHOTO 23.Unline d irrigation ditch equally as possible over his against the hillside edge,mad eb y fields. The likely systems used local Bedouin. are outlined below:

WILD FLOODING SYSTEM Thismethod , described by Booher (1974:108), consists of spilling water at frequent intervals from agrad e ditch constructed along the high edge ofa sloping field. The water isallowe d to flow freely downth e slope,irrigatin g thesoil s thewate r movesacross . Interceptor ditchesar eplace d at intervals down the slope tocollec t the water,whic h will tend toconcentrat e inswales , and to redistribute itmor e uniformly. This system isuse d primarily on steep lands or valley landswit h an uneven topography, where uniformity of water distribution isno t amajo r consideration. The system isals o important for flatter lands having shallow soils, which, for that reason, cannot be levelled. Success with this method depends on the skill toreleas e water at the correct points from the grade ditches, and in adjusting the size of openings so that the correct amount ofwate r isrelease d tocove r the area without causing soil erosion. Aminimu m of land grading isrequire d with this method of irrigation, which isa n important advantage where soil depths are shallow

123 and onlya limite d amounto feart hca nb emove d (Booher, 1974). Shallow soils coveringa ston y underground dooccu ri npart so fth eKadesh-Barne a valley.

BORDER IRRIGATION SYSTEM This method (Booher, 1974:93)require sa serie so fparalle l ridges of earth orstone , positioned ina latera l direction acrossth e valley plain. There appeart ob eancien t remainso fston y ridgesi nth eKadesh-Barne a valley which might have been used forsuc ha nirrigatio n method. Thelan di nbetwee n these parallel ridgesi scalle da border-stri p orgravity-check . The ridges guide and check ashee to fflowin g irrigation water moving downth e valley- slope, the water being released from aquaductsa tth eedg eo fth e hillside. The land should havea unifor m moderate slopean dcarefu l land preparation is necessary forthi smetho dt ob eefficient . Where conditions aresuitabl e forborde r irrigationi ti softe nth e most efficient method forth eirrigatio no fclose-growin g crops such as cereals, pastureo rothe r field crops. Thesyste m isals o used forirrigatin g orchards and vineyards (Booher, 1974).

'illiiii e

; : 6 " ' ^ ^;]^%S#!:S

PHOTO 24. A combinationo fwil d floodingan dborde r irrigation for field crops, nextt oth edate-palms , carried outb yloca l Bedouini nth evalle yo f Eine lQudeirat ,upstrea m fromth etell , southo fth estrea m channel.

124 BASIN IRRIGATION SYSTEM Thismetho d (Booher, 1974)i sgenerall y most widely used. There areman y variations in itsapplication , but the principle issimple :dividin g the land into small units so that each has anearl y level surface. (ridges, bunds or dikes)ar econstructe d around each individual unito r basin, which are filled with irrigation water to the desired depth. Basins may vary in size from 1squar e metre, used for growing vegetables and other intensive crops, to several hectares forcereal s or other field crops (Booher, 1974). The size not only depends upon the type ofcrop s but also upon the geomorphology of the landscape.Th e latter factor determines the maximum size of the basin that can be levelled.

Unlined ditches and ridges of earth are usually not preserved. In many cases these have tob e rebuilt every year on account ofnatura l deterioration or plowing. Morover,th esettin g up ofa ne w field alignment,becaus e ofcro p rotation or a change in ownership, may also alter existing field and irrigation distribution systems periodically. In archaeologic or historic reconstructions it is, therefore, often impossible torelat e hard remnants, like stone aquaducts, to soft agricultural field systems. Remains of the latter areusuall y not found. Incas e ridges arepreserved , their dating is generally problematic,henc e their correlation with remains ofknow nage . Evidence from hard but often fragmentary remnants of irrigation farming that served asth ebasi s for thisoutline , together withmor e recent historic descriptions, is presented in the following section. The position in the valley ofancien t aquaducts and damswa spointe d out toth eautho r by Yosef Porat, who investigated these remains as anarchaeologis t (Porat, personal communication, 1981).

ARCHAEOLOGIC AND HISTORIC EVIDENCE OF IRRIGATION AGRICULTURE IN THEKADESH-BARNE A VALLEY.

BRONZE AGE AQUADUCT: CARBON-14DAT E (GrN-12327) An aquaduct remnant lined with stones and limemorta r is situated at an estimated distance of about 400m downstream from the spring of Ein el Qudeirat.Th e aquaduct isburie d below the present surface of the valley plain to a depth ranging from 2.30 m for itsuppe r stones to 3.10 m for its bottom part (Photos 25,35) .Th emorta r contains small pieces ofcharcoal .

125 The radiocarbon content ofbot hth elim emorta ran d the charcoal were investigated atth eIsotop ePhysic sLaborator yo fProf . Mook (University of Groningen).Th eresultin gdate si nconventiona lC-1 4year s(BP) ,a swel la si n calibratedhistorica lyear sfo rth echarcoa ldat e(Fig .17) , area sfollows :

Mortar (GrN-12381) 12.250± 23 0 BP Charcoal (GrN-12327) 3.270+ 10 0 BP 1725-1425Ca lB.C .

Thediscrepanc ybetwee nth etw odate si sobviousl ydu et ocontaminatio n of the limemorta rwit h"dead "carbon , probablyo fEocen eage . Thelim emorta r apparently contains bothC derive d fromth eatmospher ea tth e time of the construction of theaquaduct , andEocen eC fro mth esurroundin g rocks. It seems likelytha tth eancient sproduce d limemorta rfo rth eaquaduc t by the burningo floca lEocen ecarbonat erocks :

1.CaC0 3[carbonat erock ] > CaO [quicklime ] + C02 H

2.Ca O [quicklime ] + H20 > Ca(0H)2 [slakedlime ]

3.Ca(0H) 2 + atmosphericC0 2 > CaC03 [setlim emortar ]

Either incompleteburnin go fth ecarbonat erock so rintentiona lmixin go fth e mortar withfin eroc kfragment sresulte di na C conten to fmixe d origin and agei nth emortar .Clea revidenc efo rth epresenc eo f"dead "carbo ni nmortar , probably ofEocen eage , wasobtaine db ymicroscopi cstudie so fthi nsection s fromothe raquaduc tmorta rsample si nth eKadesh-Barne avalle y (p.138) .Henc e theunrealisti cC-1 4dat eo f12.25 0+ 23 0BP , whichdoe sno treflec tth etru e ageo fth eaquaduct . Thecarbon-1 4dat eo fth echarcoa l (3.270± 10 0BP ;GrN-12327 )wa scalibra ­ ted againstth ehig hprecisio ntimescal ecalibratio ncurv eo fPearso ne t al. (1985), resultingi na historica ldat eo f1725-142 5Ca lB.C . (fig.17) . This date isadmittedl yolde rtha nexpected , whichsuggest sth eaquaduc tt o have been constructed duringth eMiddl eo rLat eBronz eAge . Thelaborator y pre- treatment of charcoalan dth ephysica lan dchemica laspect s of radiocarbon datingi ngenera lar ediscusse db yMoo kan dStreurma n(1983) .Th estratigraph y of thevalle yprofil ei nwhic hth eaquaduc tremnan tappear si s presented on page151 ,152 . Ifth echarcoa ldat ereflect sth etru eag eo fth eaquaduct ,i tseem slikel y that irrigationagricultur ewa spractise d inth eKadesh-Barne avalle y during parto fth eBronz eAge .

126 PHOTO 25. Remnantso fa Bronz eag eaquaduc t (1), aLat eByzantine-Earl y Arab aquaduct (2),bot h radiocarbon dated;an do fa recen t Bedouin pipeline(3) .

^i 1 1 1 1— 1 i i Q-

3370 JV / A

; \ /\ 3270 I ; 3170 l*^ ^\ '

iii i i i , ,iV • 1725 1600 1500 1425

HISTORICAL YEARS B.C.

FIGURE 17. Calibrationo fth eC-1 Adat eo fth e (1)aquaduc t (3270 +10 0BP ; GrN-12327)int ohistorica l years: 1725-1425Ca lB.C .

127 IRON AGE No structural remains have yet been found of Iron Age irrigation agricul­ ture. The plastered conduit, that facilitated the cistern inside the Middle and Upper Fortress tob e filled with spring water, may have been connected with a so far unknown aquaduct system upstream. The level of the conduit was about 2 metre above the IronAg e level of the valley plain in the immediate vicinity ofth e tell. Ifther e ever was such a connection, then the unknown aquaduct must have been built on pedestals or anartificia l ridge for the last hundred metre or so upstream from the fortress. The silos and granaries of the successive fortresses, as well asth e jar full of burnt wheat (Cohen, 1983)ar e ofcours e no proof of agricultural practice in the area. These discovered food storage facilities were either filled with imported grains and/or local agricultural produce. Circumstancial evidence, however, like thepositio n of the successive fortresses in the wider part of thecultivabl e valley and the occurrence of Israelite runoff farming in the adjacent highlands and various other parts of the Central Negev (Aharoni et al, 1960; Cohen, 1976), makes it illogical to assume that no agriculture was practised during the Iron Age in the well- watered valley ofEi n elQudeirat .

BYZANTINE PERIOD Woolley and Lawrence, who visited the area in January-February 1914, described remains of irrigation systems, which in their opinion belong to the Byzantine period. On both the northern and southern edge of the valley appear the abandoned ditches of two old irrigation channels against the hillside.Th e northern aquaduct is built ofmasonr y and very poor mortar. It is nearly destroyed and very difficult to trace. Maybe itwen t toth ecultivabl e valley ofWad i el Ain,whic h forms the north-western continuation ofWad i el Qudeirat in the direction ofQuseima , "but more probably itwa smad e only to bring water to a little Byzantine village whose remains yet exist in a bay of the northern slope" (Woolley and Lawrence, 1936:78,80), situated about 1k m down­ stream from the tell on the lower hillslope ofGebe l elQudeirat . Woolley and Lawrence estimated the remains at about 15t o 20smal lhouses . The aquaduct on the south side "isonl y a dug ditch, winding along the contour ofth e hills, till it comes to an end ina great Byzantine reservoir, situated in the very mouth of the valley" (Woolley and Lawrence, 1936:78), near the junction with the valley ofWad i UrnHashim . "The reservoir isa grea t work, four-square, and about twenty yards each

128 way, built in the usual style of the precise Greek masonry, laid in line,an d it still preserves inon ecorne r theopenin g ofth e sluice, which letou t its water as required to gardens on the flat land round theelbo w of the hills. The reservoir is, however, now long abandoned, half filled with earth and stones" (Woolley and Lawrence, 1936:78). More precise measurements of the reservoir were published by Elgabaly (1954): 20x 2 0x 3.5 m.Thu s the maximum capacity of the reservoir is 1400cubi c meters ofwater . It seems likely that the reservoir was used by the ancients to store the flow of spring water during the night. Irrigating at night is usually not practical for various reasons (Horst, 1983), e.g. no control of irrigation, difficult rotation between various fields, farmersar e reluctant to work during the hours ofdarkness . However, as the spring continues to flow at night, this water would be lost for agricultural production unless it is stored and subsequently released for irrigation during the daytime. The average discharge from the spring at present is6 1 cubic meters ofwate r per hour. Ina night of 12hours , during which no irrigation ispractise d by the farmers, 732cubi c meters ofwate r would flow out ofth e spring. This amount could easily be stored in the reservoir that has acapacit y of up to 1400 cubic meters. Major Jarvis was the British governor of Sinai during the years 1922-1936. He first visited the valley ofEi n elQudeira t in 1922, attracted as a sportsman by the number of desert partridges in the area. His book Desert and Delta describes his discovery ofth e reservoir, which heascribe d to the Romans, being unaware of the 1914-15 publication ofWoolle y and Lawrence who considered it of Byzantine age. Since Jarvis (1938:245) considered the Byzantine period in the as "the later Roman period, between the third and sixth centuries A.D.",ther e doesno t seem tob e a contradiction between these opinions. "Oneda y whilst resting high on the hillside overlooking the valley after a hard morning's shoot, I noticed a big square mass ofol d masonry on the southern slope ofth e hill that was obviously an old Roman reservoir. It was filled to the top with silt and the sides had fallen away inplaces , so that previously Iha d imagined it was the ruins ofa large building, but from my position above Ino w saw it for what itwas , a stone water-tank. The question then arose as toho w the Romans managed to fill it as itwa s nearly fifty feet [15 m] above the level of the valley. Gradually asm y eyes got used to the uniform greyish-yellow of the limestone rocks Itrace d the line of the ancient water-course that had brought water to the reservoir a thousand years ago and,

129 following Itu po n foot, Icam e at the head ofth e valley to the remains of the dam that had at one time raised the level of the stream." (Jarvis, 1938:244). These dam remains were pointed out to the author in 1981 by Porat, who considered them to beo f Byzan­ tine age. The dam probably had an opening in itsuppe r part to allow the raised water to flow into the very beginning of the aquaduct that is situated on top of it.Th eaqua - duct continued from the dam along the southern edge of the valley and hillside. Once it carried precious irrigation water to the fields and probably also to the reservoir, some 3k m down the valley. It is rather intriguing, however, that the position of the dam is upstream from the spring in the dry part of the valley. Woolley and Lawrence (1936:79)di d not mention the rem­ nants ofth eda m as such, but also noticed beyond the spring "a built aquaduct ofston e and lime leading out of thehill-sid e and running PHOTO 26.Remain s ofa da m and connec­ across the valley. It probably ted aquaduct in the valley ofEi n el points to another, but now forgot­ Qudeirat (upstream from the spring), ten, spring which watered these dating to the 7th century A.D. upper fields."

CARBON-14DATIN G OF THEDA M (GrN-12326) The dam and itsaquaduc t are built ofmasonry . Pieces of charcoal are usually present in the lime mortar, enabling carbon-14 dating. The author collected samples ofmorta r with charcoal from various parts of the dam on both sides of the valley to get a sufficient quantity. The charcoal was extracted from the limemorta r in the Isotope Physics Laboratory ofProf .Moo k (University of Groningen), undergoing the usual pretreatment (Mook and

130 l l i • 1 ^ CD . 1500 v\ 1470 \ -\ k _ 1380 . .

- 1290 - -

1200 -

• i 1 1 610 68S 800 900 HISTORICAL YEARSA.D .

FIGURE 18. Calibration ofth eC-1 4dat eo fth eda m (1380+ 9 0BP ; GrN-12326) intohistorica l years:610-68 5Ca lA.D .

Streurman, 1983). The extracted charcoalwa sdate d aswel la sa piec eo fmortar , taken from underneath the aquaduct surface about 110c m below thehighes t part of the dam. The resulting dates inconventiona l C-14year s (BP), as well as in calibrated historical years forth e charcoal date (Fig.18) ,ar e asfollows :

Mortar (GrN-12379) 5.330+ 130 BP Charcoal (GrN-12326) 1.380 + 90 BP 610-685Ca l A.D.

The charcoal date seemsrealistic , giving aLat eByzantin e orEarl y Arab age to theda m and itsaquaduct . The outcome forth emortar , on the other hand, doesno t reflect theag e ofth e dam,probabl y dueto ,eithe r incomplete burning by theancient so f localEocen e carbonate rock forth eproductio n of quick lime, or subsequent contamination with "dead"carbon . A microscopic study of mortar from thedam' saquaduc t not only revealed the presence of "inherited" carbonate rock fragments, but also yielded convincing evidence of thepas t flow ofwate r inth e aquaduct on topo fth edam .

131 MORTAR MICROMORPHOLOGY AND PALEO-ENVIRONMENT OF THE DAM'S AQUADUCT BOTTOM Themorta r samplewa s taken from the bottom part ofth e aquaduct,abou t 1 meter below the highest part of the dam and about 5.30 meter above the lowest part of the adjacent wadi bed. The mlcromorphological terminology used is mainly from Brewer (1964), apart from the term calciasepic, introduced by Mulders (1969). The mortar ismainl y composed of fine plasma,skeleto n grains and voids,i n decreasing order ofthei r respective volumes.Void s of various sizes appear in the mortar matrix. The size of the skeleton grains isusuall y in the range of coarse silt (0.04-0.06 mm), aswel l as very fine sand (0.06-0.1 mm). The skeleton grains are angular and subrounded in shape, appearing randomly, sometimes clustered, in the mortar matrix. Quartz grains are most common, followed by calcite, whilst hornblende, feldspars and other heavy appear occasionally. Themorta r plasma isprimaril y composed of acrysti c calcic fabric,showin g micrite to sparite calcite crystals. Thiscrysti c fabric sometimes occurs in distinct rounded grains ofu p to 1c m in diameter, which may either be rock relicts or newly formed mortar pieces with a sharp boundary. A small fossil composed of sparite proved that rock fragments containing "dead" carbon are present in themortar . Thiscause d theC-1 4dat e derived from the carbonate part of themorta r tob eunrealistic . Plasma that consists of acalciasepi c fabric, essentially made up by micrite mixed with fineclay ,als o occurs.Som e partso fth emortar , appearing as aggregates up to 1c m in size, are reminiscent ofth emicromorpholog y of loessial soil (Bruins, 1976).Man y pieces ofcharcoa l of various sizes appear randomly in themorta r matrix. The microscopic discovery ofth e following micromorphological phenomena in the mortar unequivocally proved the actual functioning of the dam and its aquaduct for a considerable time period:

AUTHIGENIC PYRITE The discovery ofa large cluster ofpyrit e framboids (photos 27, 28)i n the mortar is outstanding evidence of a reduced environment in the past, caused by a rather continuous flow of water in the aquaduct during the time of its functioning in antiquity. The microscopic identification aspyrit e became possible through the suggestions of Jongmans and Miedema (1985, personal communication). Thecluste r isu p to 0.8 mm in diameter and consists of about a hundred black toreddis h brown pyrite framboids, composed ofmicrocrystals .

132 PHOTOS 27. Incident light. PHOTO 28. Plain transmitted light. Identical microscopic pictures ofth e pyrite framboid cluster. The width of the pictures isequa l to0. 5 mm. Pyrite isopaqu e for transmitted light (Photo 28), but shows a typical shining lustre with incident light (Photo27) .

The pyrite ispartl y weathered to iron-oxide. Each pyrite framboid has a size of0.02-0.0 A mm. The pyrite cluster can be regarded asa filled crystal chamber. The pyrite framboidshav e partly penetrated the surrounding mortar matrix, showing crystal overgrowth of pyrite over micrite. For more specific aspects about the micromorphology of pyrite in soils, the reader is referred to the publications byPon s (1964), Van Dam and Pons (1972), and Miedema et al., (1974). The soil-chemical aspects of pyrite formation are discussed by Van Breemen (1972). Apart from the described cluster, more pyrite framboids occur in the mortar. Some gypsum crytals, 0.2 mm in length, may have been formed as a result ofpartia l oxidation of the pyrite into iron-(hydro)oxides. Anumbe r of ferrans and iron-oxide nodules appear in themortar .A neoferran-ferra n with a size of 0.7 mm isparticularl y well developed, partly ina vug h against the

133 voidsides and partly inside the matrix (photo 29). Such a concentration of iron- oxides is due to a rather wet environment with alternating oxi­ dation-reduction con­ ditions. It is not clear whether these ferric nodules formed in a different archae­ ological phase, cha­ PHOTO 29.Neoferran-ferra n in the mortar of the 7th racterized by a less century aquaduct on theda m (length photo is0. 9 mm; continuous flow ofwa ­ plain polarized light). ter in the aquaduct.

After the dam wasdestroyed , probably by a powerfool runoff flood inNaha l Kadesh-Barnea, the irrigation system was suddenly put out of action. This resulted in a very sharp environmental change, by which the mortar became as dry as the prevailing , except for some occasional rainfall during the winter season. As a consequence ofthi s sudden change from an aquatic to an arid environment, the pyrite has been preserved so well during the last millennium.

CARB0N-1A DATING OF AN AQUADUCT BEGINNING INTH E WADIBE D (GrN-12328) The remains ofthi s aquaduct appear in the samechanne l bluff profile, about A00 meter downstream from the main outlet ofth e spring of Ein el Qudeirat, in which the aquaduct remnant occurs (Photo 25)tha t yielded a Bronze Age C-1A dateo f 1725-1A25 Cal B.C. (GrN-12327). Whereas the latter aquaduct was a lined surface ditch, the aquaduct now under discussion had its beginning point inthi s stretch of the stream channel, tapping the water from the wadi bed (Porat, 1981,persona l communication). The aquaduct remnant consists ofwidel y spaced stones and lime mortar, which facilitated the entry of spring-water flowing over the wadi bed. The vertical dimension ofth e aquaduct is 1m , itsuppe r stones are situated ata depth ofA.1 0m an d itsbotto m part lies 5.10 mbelo w thepresen t surface of the valley plain adjacent to the stream channel (Photo25) .

13A The present level ofth ewad i bed lies at adept h of5.2 5 m and today the water from the spring flowsbelo w the aquaduct. Theaquaduc tmorta r contains small pieces ofcharcoal . The carbon-14conten t ofbot h the limemorta r and the charcoal were investigated at the Isotope PhysicsLaborator y ofProf .Moo k (University of Groningen). The resulting datesi nconventiona l C-14 years (BP), aswel l as incalibrate d historical years forth echarcoa l date,ar e as follows:

Mortar (GrN-12380) 14.250± 350 BP Charcoal (GrN-12328) 1.395 ± 50 BP 615-665Ca lA.D .

The unrealistic outcome ofth emorta r iscertainl y due to thepresenc e of "dead"carbon , for reasons described previously. Thecharcoa l date seems to represent the ageo fth e aquaduct correctly, which isvirtuall y the same as the C-14 date for theda m (1380+ 9 0BP ; 610-685Ca lA.D. ; GrN-12326). This strongly suggests that, either at theen d ofth eByzantin e period or at the beginning of the Early Arab period, two different systemswer e constructed simultaneously to tap thewate r from the spring for irrigation agriculture during the 7th century A.D.

CD 1500

1445 I CJ 1395

1345

1300

12001-

I i_J l_l i I I I Li 615 665 800 900 HISTORICAL YEARSA.D .

FIGURE 19.Calibratio n ofth eC-1 4 date of thewadi-botto m aquaduct (1395 +5 0 BP; GrN-12328)int o historical years:615-66 5Ca lA.D .

135 The position of theLat e Byzantine -Earl y Arab wadi-bottom aquaduct below the Bronze Age dated aquaduct seems an apparent stratigraphic problem. If these radiocarbon dates are correct, and the labeling ofth e two respective samples were not accidentally exchanged, either inth e field or in the labora­ tory, how is itpossibl e to find a younger aquaduct below an older one?A ke y to apossibl e answer may be thedifferen t waysb y which the two aquaducts were constructed in relation to their respective functions. The older Bronze Age dated aquaduct wasbuil t on the former valley surface to conduct the spring- water from upstream. The beginning of this aquaduct and the manner by which the spring-water entered this system isunknown . The Late Byzantine -Earl y Arab aquaduct wasno t situated at the former valley surface during the 7th century A.D., but at the level of the wadi bed to tap the flow of spring-water. Itha d to beconstructed , therefore, either by (1)diggin g achanne l from the valley-plain surface down unto a possible groundwater level adjacent to the natural stream channel, or (2)b y carving out the lowest part ofth eexistin g stream channel bank ina slight lateral excavation at the level of the wadi bed and below theBronz e Age dated aqua­ duct (ifthi s aquaduct was already present indeed): The former possibility (1)seem s unlikely for the following reasons: The stream channel ofWad i elQudeira t isth e natural aquaduct that carries the water from the spring downstream on its bed. The amount ofwate r seeping,pe r unit of time, below the wadi bed in a lateral direction isto o little to be caught in any useful quantity, if at all, by a subsurface seepage collecting type of aquaduct. As amatte r of fact,th e author made an excavation near the tell, 6m aside from thepresen t stream channel,virtuall y unto the same level as the wadi bed with its flow ofspring-water , finding the soil to be dry. Only the direct inflow ofwate r from the wadi bed into the aquaduct can tap this perennial stream ofspring-wate r effectively for agricultural purposes. Hence the aquaduct had tob ebuil t inside the 7thcentur y stream channel. It seems, however, unpractical to built such an aquaduct in the middle of the wadi bed,makin g it aneas y prey for erosion during runoff floods in the rainy season. The solution ofth e ancients, as interpreted and envisaged by the author, might have been as follows: the construction ofsuc h an aquaduct partly inside oneo fth e existing stream channel banks, by carving away only some sediment at its lowest part at the level ofth ewad i bed and leaving the upper part of the bluff intact. A low temporary dam wasperhap s built in the wadi bed,paralle l toth e aquaduct under construction, to prevent the entry of water that would havemad e the work impossible.

136 The loessialmateria lmakin g up the channel bank willno t easily collapse upon the removal of somemateria l from its lowest part. TheBronz e Age dated aquaduct remnant in theuppe r part of the bluff would,therefore ,no tnecessa ­ rily have been completely destroyed during the construction ofth e 7th century aquaduct. In fact only a small part of the Bronze age dated aquaduct has actually been preserved (Photo 35), ascompare d toth e large stretch of the 7th century wadi-bed aquaduct still insitu . The side of the latter aquaduct that faced the flow ofwate r in the wadi bed was obviously builtwit h openings inbetwee n thestones , whilst it had to be closed on all other sides,als o from above,t opreven t water losses as well as the entry of sediment that might block the flow ofwate r inside the aqua­ duct. Upon completion ofth e aquaduct, perhaps carried out in stages, its closed upper part could be covered again with some of theexcavate d sediment, so that the bluffmigh t have been restored asa relativel y firm natural wall with the water tapping aquaduct anchored inside. Somewhat farther downstream, a low threshold dam might have existed to raise the water level slightly. This would have facilitated the intake of water into the aquaduct considerably. From the point where the aquaduct became situated above the flow ofwate r in the wadi, as iteventuall y had to descend with a smaller gradient than the stream channel toemerg e onto the valley plain, its character and make-up necessarily changed from a water-intake aquaduct to amer e water conducting channel.

MORTAR MICROMORPHOLOGY OF THE 7th CENTURY WADI-BED AQUADUCT The mortar ismainl y composed of fineplasma , skeleton grains andvoids , in decreasing order ofthei r respective volumes. Voidso fvariou s sizes and shapes, like skew planes, vughs and some rare channels,appea r in themorta r matrix. The related distribution ofplasm a and skeleton grains isporphyroske - lic. The skeleton grains, angular and subrounded inshape , are mainly of coarse silt (0.04-0.06 mm)an d very fine sand size (0.06-0.1 mm), usually occuring randomly in themorta r matrix. Quartz grains aremos t common, fol­ lowed by calcite,whils t hornblende, feldspars and other heavy minerals appear occasionally. A number of large lime grains, 2-10 mm indiamete r and rounded to subroun­ ded in shape, appear inth e mortar matrix. These grains consist of a rather uniform crystic calcic or calciasepic fabric, in which virtually no skeleton grains occur.Becaus e of their rounded shape,i t isno t entirely clear whether these grains are inherited rock fragments or newly formed mortar particles.

137 PHOTO 30.A fossilan d charcoal in PHOTO 31.Charcoa l inth emorta r themorta r (photowidt h =0.3 2 mm) (photowidt h =2. 6 mm) (plain polarized light) (plainpolarize d light)

The mortar plasma has a predominant crystic calcic fabric, showing micrite to sparite calcitecrystals . Acalcia - sepic plasmic fabric also occurs.Smal l marine fossils,probabl y ofEocen eage , appear randomly in the mortar (Photo 30). Thisprove sth epresenc e ofinhe ­ rited "dead"carbonate , which caused the unrealistic C-14dat e derived from the carbonatic part of themortar . Many charcoal pieces of various sizes (0.005 -8 mm )appea r randomly scattered through the mortar matrix PHOTO 32.Pyrit e framboid (0.04mm ) (Photo 31, 32). inside charcoal (incident light)

138 AUTHIGENIC PYRITE The presence ofpyrit e framboids in the mortar, in relation to voids, providesunequivoca l evidence ofa reduction environment in thepast ,whe n the aquaduct was tapping the water flowing over the wadi bed during the Late Byzantine-Early Arab period. One distinct pyrite framboid, glittering when observed under the microscope with incident light,ha s formed inside a void of a large charcoal fragment (Photo32) . Concentrations of iron oxides (ferrans and neoferrans)als o occur in the mortar, sometimes asa weatherin g product of pyrite.Thi s seems to beindica ­ tive of aperio d with alternating wet-dry reduction-oxidation circumstances in the past. Fine gypsum needles appear in association with weathered pyrite as well.

BEDOUIN IRRIGATION FARMING INTH E KADESH-BARNEA VALLEY TheLat e Byzantine toEarl y Arab dam and aquaduct systems, described in the preceding sub-chapters, were probably destroyed by powerful runoff floods in Nahal Kadesh-Barnea during the Early Arab period. There is no evidence whether these irrigation systems were abandoned beforeo r after their apparent natural collapse. Since the Early Arab period, there seems to ben o archaeologic or historic information about irrigation agriculture in theKadesh-Barne a valley until the dawn of the 20th century. During their visit in Januari-Februari of 1914, Woolley and Lawrencemad e anumbe r ofobservation s about Bedouin irrigation farming in the valley, which appear scattered through the pages of their book (Woolley and Lawrence,1936 ,p . 75-84). At that time theBedouin' s system for irrigation agriculture involved a simple dam across the stream bed, about 800m downstrea m from the spring proper, and some 400m downstrea m from the previously described ancient aqua­ duct remnants. Thisda m was probably made ofeart h and stones. It probably had to be repaired or rebuilt regularly after the runoff floods in the rainy season. The dam raised thewate r level somewhat inth e stream channel. Here the irrigation ditches ofth e Bedouin belonging toth eGudera t tribe had their start. These unlined ditches "are only ephemeral,a deeper furrow among the crops, as such canals must always be,mad e each spring and destroyed each winter in the Arab yearly interchange ofploug h and fallow" (Woolley and Lawrence, 1936,p .78) . From this dam the valley was cultivated in a downstream direction until the

139 fields fade out at the valley mouth. The two distinguished British surveyors made an elucidating description of the hydraulic characteristics of the stream channel and its functioning during runoff floods in the rainy season, as well as during Bedouin irrigation practices in their time. "The water-course ... isa clea r channel,cu t fivet o fifteen feet [1.50- 4.50 m]int o the ground, steep banked, and generally from three to ten yards [2.70 - 9 m]i nwidth . It thus wastes only an inconsiderable part of the valley space, and itsdept h gives it content enough tocarr y off all ordinary floods without damage toth e fields on each side. An occasional great flood may sweep the whole place, levelling trees and washing out the soil, as happened two years ago [1911-1912]; but normally the lower reaches of the torrent-bed are dry except when it is raining,o r when the cultivators,havin g finished the watering ofthei r land, turn the stream ofthei r little canals back into the proper bed." (Woolley and Lawrence, 1936,p . 77-78). The British authors do not mention the growing of fruit trees by the local Bedouin, but only field crops. This iscorroborate d by Jarvis who found a similar situation in 1922, as shown ina picture in his book (Jarvis, 1938:272). The cultivated part of the valley appeared green or yellow with the crop according to the season. Acacia trees stood at intervals along the edge of the valley. Woolley and Lawrence described the tell-mound as rising above the adjacent corn-fields, where barley and/or wheat was probably growing at thetime . In front of the ruins ofth e Byzantine village, mentioned earlier, are the threshing floor and corn pits of the Guderat tribe. TheBedoui n moved their camp to the old village site for a few weeks in each year.Apar t from farming, the British surveyors also noted semi-nomadic livestock rearing as part of the food producing economy ofth e Bedouin. They mention the flocks and herds at Ain el Gudarat, while calling attention to the tribesman as a great maker of flint tools,use d for theanimals . "There is nopermanen t settlement of Arabs in the valley, through their fear of the climate. It isbelieve d that anyone who lies there in summer (whether man or beast)wil l be attacked by an intermittent fever of peculiar strength; indeed even thecooke d meat of animals which have fed in the valley is declared dangerous by the local authorities inhygiene . As a matter of fact, the large deep pools ofth e upper river must be admirable breeding places for mosquitoes" (Woolley and Lawrence, 1936,p .80) . About eight years later, in 1922, Major Jarvis first visited the area as the British Governor ofSinai ,a function inwhic h he continued to serve until

140 1936. "When I first saw the Wadi Gedeirat there were about six acres [2.4 ha] ofcultivate d land which the local Arabs irrigated bymean s of a small mud dam that they constructed every year after thewinte r floods" (Jarvis, 1938:243). This type of dam and its location was probably the same as referred to by Woolley and Lawrence,describe d above. After discovering the remnants of the ancient irrigation systems in the valley of Ein elQudeirat ,a s described on p. 129,i t occurred to Jarvis "that it would bemos t interesting and fascinating, not to say useful, to recon­ struct in some way the old Roman system of irrigation... The need for some­ thing of the sort was apparent, for the Arabs ofthi s area depend for their existence on rain crop barley which isonl y adefinit e success once in five years" (Jarvis,1938 ,p . 244-245). This remark by Jarvis isno t entirely correct, as theBedouin s were not self-sufficient in their crop farming, which constituted only an addition to their extensive livestock husbandry. The semi-nomadic Bedouin usually trade their animals for food grains grown by sedentary farmers inth e wetter parts of the region (Marx, 1982). However,th e note that rainfed barley farming in the area during the twenties and thirties was only successful once in 5 years, is a valuable datum. Jarvis (1938:245)continue d his reasoning: "Ifthe y were provided with an irrigation system that would enable them to make certain of their winter barley, and to grow also a summer crop ofmaiz e or millet, it would constitute a very definite bettering ofthei rlot" . Inspired by the ancient example, Jarvis decided tohav e a dam constructed in the stream valley for irrigation farming purposes. Funding of the dam came from the budget allotted to Sinai for anti-malaria work, as the construction of a dam would "kill twobird s with one stone" (Jarvis, 1938:247): both advancing irrigation farming and fighting malaria inth e area. The dam would hold the water back ofth eperennia l stream fed by the spring and this would prevent the formation of isolated pools downstream. "Onecoul d easily cope with themosquit o larvae inth e big pond that would form by the introduction of that jolly little Egyptian fish, the Bolti... immediately after hisintro ­ duction to a stream orpool , he makes himself at home and proceeds to breed to such an extent that in a few short months the water isa movin g mass of small fry. The fry,whic h are in aconstan t state of ravenous hunger,wor k their way up into the grassy fringes of the pools and round the stems of the reeds and rushes, so that it takes amos t redoubtable mosquito larva to escape their attentions" (Jarvis, 1938:247). The dam was constructed of cut limestone blocksan d masonry at the head of

141 the wider part ofth evalle y in 1926. Itspositio n isabou t 700m downstream from the spring, apparently near the place where theBedouin s used to make their small mud dam. Thecarbon-1 4date d ancient dam and aquaduct remnants are situated upstream from theJarvi s dam. HajAli , thegovernmen t mason at El Arish, constructed the dam, whereby he sometimes manufactured home-made lime on the spot, ifther e was a lack ofcemen t (Jarvis, 1938:248). This suggests that the ancients could also have made their lime from local Eocene carbonate rock. When the dam wascompleted , a huge pond, about 90m long and 27m wide, formed behind the dam ina day and ahalf . The barrier was not absolutely water-tight,a s someone-sixt h of the water supply was finding itswa y through thecoars e gravel beneath the concrete foundations. HajAli , therefore,con ­ structed a much lower second dam, some 45m downstream from the first. From this second dam theescapin g water was led "into the big channels that ran from theda m to thereservoirs , so that by thismean s we harnessed the whole flow of the stream" (Jarvis, 1938:249). It appeared that excellent cast-iron pipes, 20c m indiameter , and with a combined length ofnearl y 5 km, had remained in the area since the time of their use by theTurkis h army. By employing these pipes it became not only possible to run the water from the dam to the reservoir without seepage losses, but also tocarr y subsidiary channels across gorges to good land beyond, while the irrigation system was extended to about 800 m below the reservoir at themout h ofth e valley (Jarvis, 1938:250-251). "The reservoir was then cleaned of the silt of ages,whic h filled it to the brim, and the old Roman plaster lining,whic h had fallen away in patches,wa s stripped off and a layer of twentieth-century concrete substituted. Thecon ­ crete was later faced with a lining ofcemen t and 'Pudlo', a patent water­ proofing mixture, and the reservoir was ready for the flow from the stream" (Jarvis, 1938:251). TheBritis h governor admitted that he "felt something ofa Philistine" when he lined the age old construction "ofvas t cut stones with the prosaic and unpleasant-coloured concrete ofto-day , but Iconsole d myself with the thought ofth e acres of dun-coloured deserts that would become green as the result." Ittoo k a day to fill thereservoi r ofabou t 20mete r square and 3.5 mdeep , which fits thecalculatio n made on page 127, in relation to the average discharge ofth e spring of6 1 cubic meter per hour. Jarvis described allkin d of social and human-psychological problems which had to be overcome, aswel l as the need for instruction, before the potential ofth e irrigation system was tapped by the local Bedouin population. A govern-

142 ment garden was established, both as an object lesson and for experimenting with various crops and fruit trees. The olive treesmad e the most prolific growth,attainin g the height of over 3.50 meter inthei r third year.Th e vines were also exceptional inever y way, producing a remarkable quantity of grapes with a very fine flavour. Citrus trees proved less successful. Concerning vegetables,th e growth ofasparagu s wasphenomena l inever y respect. The irrigation scheme in the valley ofEi n elQudeira t was started by Jarvis in 1926 and itha d been inexistenc e for ten yearswhe n he left Sinai in 1936. Some 120hectar e had come into cultivation with thousands of olive trees in varying stages ofgrowth . "As Iexpected , there wasn o realenthou - siasm for the work until the first Beduin had marketed hiscro p of olives and turned the cash proceeds over wonderingly inhi s palm. Then, when it trans­ pired that there was real money to be made over these queer bitter fruits, every man proceeded to extend his plot, and to level off fresh desert land until finally one found irrigated orchards amil e away from the dam" (Jarvis, 1938:268). Apart from olives, the bedouins also cultivated vines, winter grains,an d somemaiz e insummer . Some twenty years later, in the fifties, the system still seemed to function properly, according to Elgabaly (1954)wh onote d that the water was carried in iron pipes and cement canals to irrigate theare a below the dam. During the night water wascollecte d and stored ina reservoi r with a capacity of about 1000cubi c meters. He reported orchards ofolives , pomegranates and figs growing on the water from the spring, and considered the olives as the most successful trees. What Jarvis (1938:269) had feared that "theda m may crack with a heavy flood" happened in 1956, thirty years after its construction. However, the repaired the dam, which, as a result, continued to function until 1969,whe n another powerfool runoff flood broke the dam again (Fromkin, 1981). Since that event, theJarvi s dam has not been reconstructed (Photo 33), and the local Bedouin farmers use the cast-iron pipes tota p the water from the spring in order to irrigate their orchards and field crops (Photo 24). The remaining structure ofth eJarvi s dam functions today asa kin d of aquaduct- bridge for the cast-iron pipes to cross the valley (Photo33) . In a talk with a local Bedouin inDecembe r 1981, Iaske d whether they manure their fields. "Only water" was his answer, ash e denied any use of whatever kind of . The and sheep drop,nevertheless ,a certain amount of manure, as they are allowed to graze on uncultivated plots and various other fields.

143 10 Kadesh-Barnea: valley stratigraphy and soil development

THEVALLE Y SECTION NEARTH E LATE BYZANTINE- EARL Y ARABDA M

Historic reconstruction oflandscap ean dsoi l development inth e Kadesh-Barnea valley became possible through thepresenc eo f archaeological remainsan dradiocarbo n dating.Th evalle y sections which supplied useful data are dealt withi norde ro fthei r respective locationi nth e stream channel, starting with theC-1 4date d Late Byzantinet oEarly-Ara b dam,upstrea mo fth e present positiono fth e spring,an dsuccessivel y continuing downth evalley . Remnantso fth eda mappea ro nbot h sideso fth e stream channel, asca nb e seeno nPhot o 26. Acros s sectiono fth e valley (Figure 20)show sth e strati- graphic positiono fth eda mremain si nrelatio nt oth e accumulated sediment and present levelo fth e stream bed. Iti seviden t from this stratigraphic relationship that,som e time afterth e destructiono fth edam , sediments were deposited inth e valley duringa remarkabl e periodo faggradation , followedb y a periodo fincisio n that seemst ocontinu e even today.Thi s relatively recent periodo fdowncuttin g exposed the above sedimentary profile for examination. Some sediment samples were takenfo rlaborator y analysis from thisprofile , which isprecisel y locatedi nbetwee n the tworemnant so fth e7t hcentur yda m (Photo 26, Figure 20), situateda topposit e sideso fth e valley. In other words, once theda mi nit sful l length blocked the valleya tthi s spot where now sediments appeari nit splace . Most ofth e sediments exhibited inth estudie d profilear e rather uniform in texture, beingo fa loess-lik e nature without macro-structural elements, untila dept ho f3.9 5 m, whena lithologica l break appears. The baselevelo f the LateByzantine-Earl y Arab damcoincide s witha dept ho fabou t 2.55 m.

144 PHOTO 33.Th e brokenJarvi sDa ma sa moder n exampleho wth e 7thcentur yda m might have looked after itspartia lcollaps ean dbefor eth e unique historic periodo fwad i aggradation,durin g whichth eda mbecam enearl y buried.

f ,r7T^r,i,i,i,1,1,1, 1 iI,;, ! DAM & AQUADUCT

205c m

WADI BED TODAY

FIGURE 20. Cross-sectiono fth estrea m channeli nrelatio nt oth eremain s of the 7thcentur yda mi nth evalle yo fEi ne lQudeirat .

It seems likely thatth e entire alluvial loessial sequence appearing aboveth e lithological change at3.9 5 mpostdate sth e destructiono fth e 7th century dam,deposite d duringth e periodo fextraordinar y aggradation. This stratigraphic conclusion impliesa phas eo fdowncuttin gan d erosion, following the destructiono fth eda mbu tprio rt oth e aggradational period, which removed some 1.50 mo fsediment s belowth e base levelo fth edam , froma

•145 SURFACE level ofabou t2.5 0unti la dept ho f3.9 5m . ALLUVIAL DEPOSITS Many dark organiclayer s occur in the 30 sediment profilea tinterval so fabou t1 0t o 50 20 cm. Mosto fthes edar klayer sar e faint andthin , lesstha n1 c mi nthickness .Thre e moreprominen tlayer swer edistinguishe d ata 140 depth of 1.40, 2.05an d3.2 0m below the surfaceedg eo fth echanne lbluff . The dark 205 layers musthav ebee nancien tsurfac elevel s 220 240 (paleoA-horizons) ,onc ecovere dwit hvegeta ­ -BAS E LEVEL DAM 255 tion, apparently fed by water from the spring.Thes esurfac ehorizon swer esuccessi ­ 320 vely formedan dsubsequentl y buried during the gradual processo fvalle y aggradation. 395 Today vegetationmat scove r little surface 420 areas within the streamchanne l from the 440 460 ^mro^f^ first emergenceo fsprin gwate ri nth e wadi PRESENT WADI BED bed. X Sample Organic layer Afe whundre dmeter sdow nth evalley ,som e '. Distinct gley spots 50 m upstream fromth emai n oulet of the spring proper, Goldberg (1984)wa sabl e to FIGURE21 .Sample dprofil ei n collect enoughcharcoa lfo rcarbon-1 4datin g betweenth eda mremains . from a pocketo rlens ea ta dept h of 3 m belowth evalle yplain . Thecharcoa lyielde d a date of 665 +11 5B Pi n conventional C-14 years. Calibrated against Pearson'scurv e(Pearson ,1986) ,thi sdat ecorrespond st oa historica ldat eo f 1260-1400Ca lA.D . (Figure22) . Thelowermos tprominen torgani claye ri nth e alluvialsediments , situatedbetwee nth e7t hcentur yda mremains ,occur sa ta similardept ho f3.2 0m an di spresume dt ob eo fidentica lage . Ata dept ho f3.9 5m a lithologica lchang ei svisible ,marke db yver yclea r -browngle yspots ,indicativ eo falternatin goxidation-reductio ncircum ­ stances in thepast , duet oaquati ccondition si nthi spar to f the stream channel.Anothe rdistinc tlithologica lchang eoccur sa ta dept ho f4.2 0m ,th e upperboundar yo fa mor esand ylayer , 20c mthick ,tha tcontain sman ymollus c shells.A t4.4 0m a ston ylaye rappear sa tth ebas eo fth eprofile ,exhibitin g subroundedstone su pt o3 0c mi ndiameter . Theaverag esiz eo fth estone s is 10-20cm ,whil eth einterlockin gpackin gvoid sar efille dwit hsmalle rgravel . Thestone sar eimbricate di na downstrea mdirection .

146 1 0_ CD v \

<: 780

i 700 i\ 665 1 \ / \ 600 \J\ 550 \

i i .! . . "L. 1200 1260 1400 HISTORICAL YEARSA.D .

FIGURE 22. Calibration of theC-1 4dat e ofa charcoa l sample (665± 115 BP; Goldberg, 1984), collected from adept h of3 m belo w the valley plain, into historical years: 1260-1400Ca lA.D .

The lithological change at 3.95 or 4.20 can beregarde d as an erosional discontinuity, that marks the depth ofwad i incision after the destruction of the dam and prior toth e following period ofaggradation . Theabov e mentioned carbon-14date , being younger than the age ofth e 7th century dam, fits this picture very well indeed. It thus seems likely, that by the 13th or 14th century A.D., thewad iha d already deposited its first 75-100 cm of alluvial sediments: the onset of theuniqu e historic period of wadi aggradation. Another 3m o fsedimen t was yet to bedeposite d inth e following centuries. Two sediment samples were taken inth ealluvia l loessial deposits at respective depths of30-5 0c m and 220-240cm . Another sample was collected from themor e sandy layer at 420-440cm , which containsmollus c shells. The laboratory results aredisplaye d inTabl e 18. Thealluvia l loessial sediments are rich in carbonates and appear tob emoderatel y saline throughout the profile. Sodium and chloride make up the bulk ofth e salts. The amount of sulphate isrelativel y high inth e layer near thebas eo fth e section, which also shows acomparativ e decrease inchloride .Th ehighe r nitrate value inth e middle ofth eprofile , may have been derived from someo fth edar k coloured organic layers.

147 TABLE18 . Granulometryan dchemica lanalysi so falluvia lsediment si nbetwee n the remnants of the radiocarbondate d7t hcentur y dam in the valleyo fEi ne lQudeirat .

DEPTH CLAY SILT SAND TEXTURE GRAVEL CARBONATES <0.002 0.002- 0.063- >2m m CaC03 mm 0.063 2 mm (cm) (») (S) (K) (S) (8)

30-5 0 15.8 44.7 39.5 LOAM 0 43.6 220-240 10.4 38.6 51.1 LOAM 0 44.3 420-440 17.1 19.6 63.4 SANDYLOA M 1 56.5

DEPTH PH EC(1:1) SAR ORGANIC TOTAL CARBON PHOSPHORUS (cm) (mS/cm) (»)c (55) P205 30-5 0 7.1 9.5 49.0 0.70 0.13 220-240 7.2 9.3 49.4 0.27 0.14 420-440 7.4 8.8 71.6 0.36 0.09

DEPTH SOLUBLESALT SI N 1:1SOIL/WATE REXTRAC T (meq/kgsoil )

2+ (cm) Cra Mg2+ Na+ K+ HC0~ Cl" SO^" NO3

30-5 0 6.1 0.2 87 0 0 1.0 91.0 5.0 0 220-240 6.0 0.2 87 0 0 0.9 89.5 5.0 1.2 420-440 1.6 <0.1 65 0 0 0.4 35.0 33.0 <0.1

Percentagesi nweigh t %o fth efin eeart h< 2mm ; onlygrave li nweigh t % of thetota lsoil . EC= electrica lconductivity . SAR= sodiu madsorptio nratio .

WATER INLET INTO

AQUADUCT DAM TthCENTORV A.D.

BED LEVEL 38 m WADI

VTAVB^D LEVEL

FIRST APPEARANCE OF SPRING WATER TODAY

FIGURE 23. Longitudinalvalle ysectio nthroug hth e7t hcentur ydam , showing pastan dpresen thydro-geomorphi crelationships .

148 Figure 23 shows the location of the dam with respect toth e present posi­ tion of thewate r level from the spring, aswel l as the likely situation in antiquity, when a pond must have accumulated behind thedam , so that the spring-water could flow into the aquaduct on top ofth e dam. Theheigh t of the dam was about 4.30 m. From the studied stratigraphic interrelationships the following conclusions seem warranted about theevolutio n ofthi spar t of the stream channel inhistorica l times:

1. The level of thewad i bed at the time ofth edams' s construction (7th century A.D.)wa sabou t 2mete r higher than its lowermost present level. 2. The nearest and highest occurrence of spring-water today appears at a distance of 38m downstrea m from the dam, 3.30 mbelo w thebase-leve l of the 7th century dam. 3. Ifth e dam was tob ereconstructe d today, it would serve no other purpose than the temporary retention of some runoffwate r during floods in the rainy season. 4. The apparent function ofth e dam, in antiquity, wast orais e the waters from the spring to the level ofth e aquaduct system that began on top of the dam, for irrigation farming purposes. 5. Hence an important outlet of the spring was inal l likelihood situated upstream from the dam in the 7th century A.D.Ther e isnon e today. 6. At some time during orafte r the 7th century A.D.th eda m was destroyed by a powerfool runoff flood inNaha lKadesh-Barnea , comparable to the des­ truction ofth e Jarvis dam in the 20th century. 7. The destruction ofth e dam was followed, during acertai n period oftime , by a regime of stream channel incision. 8. This period ofdowncuttin g was terminated by a remarkable change in the hydraulic and geomorphic behaviour ofNaha l Kadesh-Barnea. Sediments began to aggradate in this part of the stream channel, as aresul t ofwhic h the position of thewad i bed became higher and higher. Eventually, the rem­ nants of the 7th century dam were almost completely buried by alluvial loessial sediments. 9. Another environmental change caused the cessation ofthi s quite extraordi­ nary period ofwad i aggradation. Nahal Kadesh-Barnea began tocu t into its own bed, removing a considerable part of its previous deposits. This period of incision is still continuing today and has advanced to such a degree that thepresen t level of thewad i bed iswel l below the level of the 7th century A.D.

149 It thus seems that the position ofth e spring in the ^gs|£*r *y ^ .S^ffiS/';&aK ,•-2A^ >^ . valley ofEi n elQudeira t has shifted since the 7th century A.D. Concerning the actual siteo fth e spring today,on e has to beawar e ofth e two manners inwhic h the spring- water emerges inth e stream channel. The first quiet appearance of spring-water in the wadi bed begins 38m downstream from the location of the 7th century dam.Thi sver y gentle emergence and quiet flow of water in thewad i bedconti ­ nues fora fewhundre d meter furtherdow n the streamchan ­ nel, until themajo r present outlet of the spring is reached (Photo34) . Hereth ewate r risesnatu ­ rally with some force through PHOTO 34. (1)Locatio n ofth e 7th century a concretepipe , vertically dam. (2)Th e firstemergenc e ofspring - placed into the wadi bed. water in thewadi-bed . (3)Th emai n outlet This "concrete"sigh t of the of the spring atpresent . spring is,admittedly , rather disappointing today,compare d with itsappearanc e in 1914, asdescribe d byWoolle y and Lawrence (1936:79): "Thesprin g proper is...wher e abuttres s oflimeston e runs into thevalley.. . From the footo fi tth ewate r gushesou t strongly inthre e little spoutsthic k as aman' s arm, from deep,narro w fissures inth e rock...Ai n elGudera t means the spring ofth eearthenwar e kettles, or small spouted pots. Whether it refers toth e rush ofwater ,i ncontras t toth e slowwellin g up of AinKadeis , or toactua l pottery,w ekno wnot" . Since the reference toth e limestone buttress isunclea r today,on e wonders whether theprincipa l outlet of the spring hasactuall y changed since 1914?

150 THECHANNE L BLUFF PROFILE THAT EXHIBITS TWOC-1 4DATE D AQUADUCT REMNANTS (BRONZEAG E AND LATE BYZANTINE-EARLY ARAB PERIOD)

Two different aquaduct systems appear inth e southern bluff of the stream channel,abou t 400mete r downstream from themai n outlet of the spring. These remnants were radiocarbon dated by Mook at the University of Groningen (GrN-12327, GrN-12328), and have already been described (p. 134).Th e lower aquaduct near the base of the profile, situated somewhat above the present level ofth ewad i bed, formed the beginning of awate r transport system in the 7th century A.D. Thisstretc h of aquaduct had the function to tap the spring- water flowing over thewad i bed. The remains of theaquaduc t are still visible over some tens of meters (Photo 35). Itwa s built at the edge of the stream-channel bottom existing at the time and itspositio n indicates that thewad i bed of the 7th century A.D. was, at this part ofth e valley, perhaps onemete r higher than at present. This conclusion fitsth e evidence from the 7th century dam site, some 700 m upstream, which also indicated ahighe r level of thewad i bed in thosetimes .

PHOTO 35. Apartl y collapsed remnant of aBronz e agedate d aquaduct (1)abov e extensive remains ofa 7t h century (A.D.)aquaduc t (2)nea r the wadi bed.

151 SURFACE The aquaduct remnanti n themiddl e ALLUVIAL DEPOSITS part ofth e profile, ofwhic h only a

45 small parti sstil l preserved (Photo35) , 55 yielded a surprisingly early carbon-14 date (GrN-12327)o f1725-142 5 Cal B.C. (Late-Middlet oLat e Bronze Age).I ti sa

160 - typeo faquaduc t that probably was situa­ 180 ted at theforme r surface levelo f the valley duringth e timeo fit s functioning 230 AQUADUCT inth e past (Porat,1981 ,persona l commu­ 270 290 [1 1725-1425 Cal BC nication). Theuppe r 2.30t o3.0 0 meter 310 of sediment above this aquaduct (Photo 35)must , therefore,hav e been deposited 350 Y 370 - since theLat e Bronze Age duringa perio d

410 of wadi aggredation. AQUADUCT The alluvial sedimentso fth e entire 615 -665 Cal AD profile (5.25m )ar eo fa loess-like 510 nature, consisting ofsil t loamt ocla y 525 PRESENT WATER LEVEL loam inth elowe r parto f thesection . AND WADI BED The deposits havea somewha t finer tex­ X Sample ture than the loamy sediments nearth e FIGURE 24.Sedimentar y profile 7th century dam, some 700m farther up­ of the channel bluff,i nwhic h stream. Sediment samples were takena t the ancient aquaducts appear. respective depthso f45-5 5 cm,160-18 0 cm, 270-290 cm (inside thesediment - filled Bronze Age dated aquaduct),an d350-37 0c m(i nbetwee nth etw o aquaduct remains).Th elaborator y results arepresente di nTabl e19 . The clay contento f7 %insid eth e Bronzeag e dated aquaducti sth elowes t value encountered inth e investigated valley sediments. Thetw osample s from the alluvial sediments overlying theBronz eag e dated aquaduct remnant, appa­ rently deposited during themai n periodo fhistori c wadi aggradation, havea somewhat different granulometry thanth edeepe r layers. Thelaye r aboveth e wadi-bottom aquaducti slikel yt ohav e been disturbedb yma ndurin g the con­ structiono fth e latter aquaducti nth e7t h century A.D. The entire sediment profilei smoderatel y affectedb ysalinity , which hasa tendency toincreas e slightly ina nupwar d direction. Thesam e trendi s also perceptiblei nth e sediment profilei nbetwee nth e remnantso fth e7t hcentur y dam. This kind ofdistributio no fsalt s through theprofil e suggests their

152 TABLE 19. Granulometry andchemica l analysiso falluvia l sediments in the channel bluff profile that exhibits the C-14 dated aquaduct remnants (* 1725-1425 CalB.C . and# 615-66 5 CalA.D. ) @ inth evalle y ofEi ne lQudeirat .

DEPTH CLAY SILT SAND TEXTURE GRAVEL CARBONATES <0.002 0.002- 0.063- >2m m CaC03 mm 0.063 2 mm (cm) (%) (») (») (S) («)

45-5 5 20.2 58.9 21.2 SILT LOAM 0 41.7 160-180 23.9 59.6 16.5 SILT LOAM 0 44.9 * 270-290 7.1 67.4 25.4 SILT LOAM 0 38.1 350-370 27.3 35.2 37.5 CLAY LOAM 2 46.9 #

DEPTH pH EC(1:1) SAR ORGANIC TOTAL CARBON PHOSPHORUS

(cm) (mS/cm) (8) C (,%) P2O5

45-5 5 7.3 10.4 92 6 0.5 2 0.12 160-180 7.3 12.8 145. 0 0.5 2 0.14 * 270-290 7.3 11.2 99 2 0.5 4 0.14 350-370 7.4 9.9 56. 5 0.5 2 0.11 #

DEPTH SOLUBLESALT SI N L:lSOIL/WATE R EXTRACT (meq/kgsoil )

2 + 2 (cm) Cra Mg2+ Na K+ HCO3 CI" SO " NO3

45-5 5 3.7 0.35 131 0 <0.1 0.50 107.0 27.0 0.6 160-180 1.7 0.30 145 0 0 0.20 133.0 11.0 0.3 * 270-290 5.8 0.35 174 0 0 1.15 134.0 46.0 0.1 350-370 4.5 0.25 87 0 0 0.40 54.0 39.5 0.2 #

Percentages inweigh t %o fth efin e earth <2mm . Only gravel inweigh t %o f the total soil. EC= electrica l conductivity. SAR= sodiu m adsorption ratio. (8Not e peculiar stratigraphy: younger aquaduct below older remnant (p.134).

origin tob erelate d toa capillar y riseo fspring-wate r from the wadi bed (Kamphorst, 1986,persona l communication;Kamphors tan dBolt , 1978). However, the levelo fth ewad ibed , andwit h itth eleve lo fth eflo wo fspring-water , hasexperience d dramatic changes sinceth eByzantin e period. Thesal tdistri ­ bution, therefore, cannotb erelate d toByzantin eo rearlie r periods. Salts are easily leached andthei r present distribution inth esediment s adjacentt o the stream channel probably reflectsth emor e recent local environmental history,i.e .th emai nperio d ofwad i aggradationan dth esubsequen t periodo f downcutting.

153 The amount ofdissolve d salts in the spring water isabou t 1.42 g/liter. Sodium and chloride are the main salt constituents in the spring-water as well as in the sediment.Bot h of these components decrease inquantit y in the lower part of the sediment profile, accompanied with an increase in sulphate. A similar salt distribution occurs in the sediment profile in between the 7th century dam remnants,som e 700m upstream.

VALLEY AGGRADATION AND ITSRADIOCARBO N DATING NEAR THE TELL

Going downstream from thechanne l bluff site that exhibits the remains of the two ancient aquaducts, one passesth e partly destroyed Jarvis dam after about 250m .Th e tell of the fortresses lies further down the valley at a distance of some 1200m from the former aquaduct site.

PHOTO 36. A large stretch of the valley ofEi n el Qudeirat, displaying the location of: (1)th echanne l bluff site with the ancient aquaducts, (2) the broken Jarvis Dam, (3)th e tell of the fortresses.

154 About 200 mupstrea m fromth etell , acharcoa l pocketwa sfoun d in the southern channel bluff,approximatel y 2.50m belowth evalle y plain and1.5 0m aboveth epresen t streambe dwit h itsperennia l flowo fspring-water . Achar ­ coal samplewa stake n (GrN-11946)fo rcarbon-1 4dating . Some 60m southwest ofth etell , another charcoal sample (GrN-11947) was collected from thenorther n channel bluff, ata leve lo fabou t 3m belo wth e valley plainan d1.3 0m abov eth epresen t streambe dwit h itsgentl e flow of water. Both charcoal samples were investigated atth eIsotop e Physics Labora­ tory ofProf . Mook (University ofGroningen) . Theresultin g dates inconven ­ tionalC-1 4year san dcalibrate d historical years (Figure25 )ar ea sfollows :

GrN-11946 540+ 4 5B P 1310-1430Ca lA.D . GrN-11947 445± 2 5B P 1440-1460Ca l A.D.

CD

Ld

585

i \ 540 ~ i | i l i V 495 475 _._._. J.._._._. _ i\ 445 i :7\ 415 1200 1300 1400 1500 HISTORICAL YEARSA.D .

FIGURE 25. Calibration ofth eC-1 4date so ftw ocharcoa l samples, collected from different channel bluff sections nearth etell , into historical years.

These data give important information about wadi aggradation during the Mamluk Period, inadditio n toth edetaile d picture already obtained from the 7th century damsite , upstream fromth espring . However,mor e informationi s yet needed to establish more precisely theduratio n of the extraordinary period ofaggradatio n after.Byzantine-Earl y Arab times.Moreover ,i sther ean y available evidence about stream channel evolution inth evalle y of Ein el Qudeirat during earlier historic periods?

155 FIGURE 26. Stratigraphy of valley deposits atth e southern edge of the tell.

156 As thewad i flowsalon g the southern edge ofth e tell, itwa sassume d by the author that thispar t of the valley might contain stratigraphic infor­ mation about the position and evolution of the stream channel and itswad i bed during theLat e Bronze, Israelite and subsequent periods.Therefore ,a trench was dug from the base ofth e southern buttress wall in the direction of the stream channel inth ewinte r of 1981/1982, during thelas t seasons ofexcava ­ tions by Rudolph Cohen before the Israeli withdrawal fromSinai . Itwa s discovered that aterrace d slope, lined with large stones,descend s from the buttress wall ofth e Middle and Upper Fortress toward the former position of thewad i bed during the Iron Age. The obvious purpose of this terraced slope, covered with stones, was toprotec t theancien t Israelite fortresses from being undercut by powerfool runoff floods inth e rainy season. The stratigraphic relationships between the terraced slope, the various dark ash layers, accumulated debris and sedimentary layers deposited by the wadi at this part of the tell,ar e shown inFigur e26 . The baseo f thebuttres s wall lies2 mete r higher than thepresen t level of the stream bed. The distance between thebuttres s wall and the stream channel is 12m ,whe n measured along the trench. The terraced and stone covered slope, descending from thebas eo f the buttress wall in the direction ofth e wadi, marked the landscape surface between the tell and the stream channel during the Iron Age. Anatura l gravel deposit, probably ofLat ePleistocen e age, underlies the base ofth ebuttres s wall and the first part of the terraced slope. The position of the stream channel during the Iron Agewa s similar as today, but the level of the wadi bed might have been somewhat lower than at present. Alluvial loessial deposits cover the terraced Iron Age slope and appear unto aheigh t of about 4mete r above the level ofth epresen t wadi bed. A sloping dark ash layer, about 5mete r in length, descends from the buttress wall,passe s justove r the top ofth e firstman-mad e Iron Age terrace step, and ends within the alluvial loessial sediment at aheigh t of 1.50 m above the present stream bed level. This dark ash layer marks the landscape surface at atim e when thewad i aggradation had already begun, since at least some 1.50 mo fsedimen t was already deposited. Three charcoal samples from different parts ofthi sas h layer were dated atth e Isotope Physics Laboratory ofProf .Moo k (University of Groningen). The resulting dates in conventional carbon-14year s aswel l as in histori­ cal or dendro years, after calibration with thecurv e of Pearson et al. (1985), are as follows (Figure27) :

157 GrN-11949 395± 4 5B P 1450- 148 5Ca lA.D . GrN-12331 340+ 4 0B P 1470- 153 0Ca lA.D . GrN-12333 410+ 3 5B P 1450- 148 0Ca lA.D .

1 —i r i 1 T"

CD 450 "Nt < 410 ( _ h 395 *•

I c «\ ^ IK ./">» 1. •M XI f \A 340 h 1: ™ , «: 295 .^— . . A i\ 1 i i; • i i i, \ , 1400 1500 1600 1700 HISTORICALYEAR SA.D .

FIGURE27 .Calibratio no fth eabov eC-1 4date sint ohistorica lyears .

Theactua lcalibratio no fth ethre eC-1 4date si sshow ni nFigur e27 ,whic h exhibits theparticula rswing so fth ehig hprecisio ncurv eo fPearso n (1986) in thisparticula rtim estretch . Theinclusio no fth eyounges tswin go f the graphint oth ecalibratio nseem sunrealisti can dhas ,therefore ,bee nomitted , asth edate ssho wa distinc ttendenc yt obelon gt oth eearlie rtim esectio no f thegraph .A clea rconsensu sresult sfro mth ethre edifferen tcharcoa lsample s of this dark ashlayer . Anag eo fabou t147 0Ca lA.D . appearst o be the weighedaverag eoutcome , datingthi sancien tsurfac elaye rt oth elates tpar t ofth eMamlu kPeriod . Anotherdar kas hlense , only4 0c mlong , issituate dagains tth ebuttres s wall aboveth eforme ras hlaye ra ta ver yimportan t stratigraphic position. The lense liessom e3 0c mabov eth euppe rpar to fth eforme ras h layer and some3 0c mbelo wth ehighes taggradatio nleve lo falluvia lloessia lsediments . Charcoal fromth elens ewa sinvestigate db yth eIsotop ePhysic sLaborator yo f Prof. Mook (Universityo fGroningen) .Th eresultin gdat ei nconventiona lC-1 4 years and incalibrate dhistorica lyears , accordingt oth e high precision curveo fPearso n (1986),i sa sfollows :

158 GrN-12332 70 + 170B P 1665 -Recen t CalA.D .

—r T 1 1 CD 240 r T 13- r

200

I i\ A • 100 70 - il/v •

1 1 1 ! L i i i i 1600 1665 1710 1800 1900

HISTORICAL YEARS A.D.

FIGURE 28.Calibratio n of the above C-14 date into historical years.

This result indicates that the wadi aggradation was probably at its highest level during the 17th and 18th century, when thewad i bed was about 4 meter above itspresen t level. As amatte r of fact, an abandoned stream channel exists just west of the tell, exhibiting apaleo-wad ibe d at a level of4.3 3 m higher than the adjacent present stream bed. This abandoned stream channel branches offth e present wadi course near the southwestern corner ofth e tell, some 5m south ofth e southern buttress wall, and bends away from the present stream channel before rejoining it some 200m farther toth ewes t (Photo 37 and38) . The gravel lensesappearin g in theuppe r part ofth ealluvia l sediments in the studied section (Figure 26), lie at adistanc e of4 t o6 mete r south of the buttress wall. These gravel lenses show how thewad i bed moved gradually closer to the tell during itsupwar d aggradation, ascompare d to its position during the Iron Age and today. When did the period of aggradation come to anend ? The former C-14 date (GrN-12332) indicates this tohav e occurred after the 17th century. From the observations ofWoolle y and Lawrence (1936:77)i t isclea r that the downward incision of the wadi and the formation of thepresen t stream bed had already largely taken place before 1914. It seems, therefore, likely that this remar­ kable period of stream channel aggradation came to anen d inth e 18th century.

159 m

*• ••»" iJ -JL mm. * I •• .

^iJasj!*^**

aw: *4ta 'Mt-'tr-iaBr: *,;?l^sft*is £ !!**&*

PHOTO 37. The present stream channel (1)an dth e abandoned paleo-wadibe d(2 ) downstream from the tell, which functioned during the main aggradational period. The level ofth e paleo-wadibe di s 4.33 mhighe r than the present stream bed.Not e theancien t man-made wallo nGebe le lQudeira tt oth e north. SB

,%^2fe'A \ i.'y™.''•"' ' .'•/.:,

it ^*m%, Mg-^" f„,r^- ••

PHOTO 38.Close-u p ofth e abandoned paleo-wadibe d (2), which functioned during the main aggradational period; viewed from Gebele l Qudeirat. 160 Very striking isth e sudden change in the hydraulic or geomorphic behaviour ofNaha lKadesh-Barne a from aggradation to incision atth een d of this period, aswel l as the rapid rate ofdowncuttin g that followed thismarke d environmen­ talchange .

SOIL DEVELOPMENT If aggradation took place in the stream channel during the Mamluk and Ottoman periods, by which thewad i reached a level ofabou t 4 meter higher than today, then the flow of spring-water carried by the wadi bed from farther upstream, must also have streamed up to about 4mete r higher than today. Is there any evidence of these former wet conditions inth e sediment asa result of post-depositionalprocesse s of soil development, proving wet conditions in the past? The answer isa ver y definite yes. Thin sections of three samples from the alluvial loessial sediments, depo­ sited during the period of wadi aggradation, were investigated under the microscope. Sample KB-129wa s collected from the uppermost, sandwiched allu­ vial loess layer (Figure 26), representing thehighes t level of aggradation during the 17th or 18th century. The other two samples are from somewhat greater depth. The approximate ages of the micromorphology samples,a swel l as their precise position with respect to the IronAg e buttress wall and the present wadi bed, are shown in Table20 .

TABLE 20. Position and approximate age ofalluvia l loessial sediment samples in the Kadesh-Barnea valley, south of the tell, investigated microscopically from thin sections, inorde r to study their post- depositional soil development.

Sample Depth Height above Distance Approximate time (cm) base-level of South of North of of sedimentation the Iron Age buttress present buttress wall wall wadi-bed (cm) (cm) (cm)

KB-129 30-6 0 140-170 125 1050 17-18th century A.D. KB-131 120-140 60- 80 350 825 16-17th century A.D. KB-132 190-220 10- 0- -(-)20 300 875 15-16th century A.D.

161 MICROMORPHOLOGYAN DPALEO-ENVIRONMEN TO F KB-129 The general appearance ofth ematri x isrelativel y loose, which is not surprising for such ayoun g sediment, less than 300year s old,nea r the valley surface. Three slaking crusts in the thin section areapparentl y remnants of sedimentary layering from the time ofdeposition , which seems to have been largely destroyed within aremarkabl e short period by biological pedotubation. The soil matrix iscompose d of fine plasma and skeleton grains inabou t equal volume amounts, aswel l as voids. The related distribution between skeleton grains and plasma varies, being intertextic, agglomeroplasmic,o r porpyhyro- skelic.Th e plasma hasa calciasepi c to very finecrysti c calcic fabric. The skeleton grains, angular and subrounded in shape, are generally of coarse silt to very fine sand size (0.04-0.1 mm); some sand grains are up to 0.3 mm. The arrangement ofth e grains is random, sometimes clustered. Their mineralogy isdominate d by quartz, followed by calcite, whilst feldspars, hornblende, glauconite and other heavy minerals appear occasionally. Remains of fossils,fro m silt to sand sizes, are very common.Mos t of these small fossils appear to be ofmarin e origin, apparently derived from the omnipresent calcareous marine rocks in the catchment ofNaha l Kadesh-Barnea. Charcoal remnants, varying from silt size up to 1mm , occur randomly in the matrix. A few angular to subrounded limestone fragments, up to 3.2 mm in diameter,ar e the largest components inside the sediment sample,togethe r with a piece ofpotter y ofsimila r size. Voids of various sizes and shapes, like channels, compound packing voids, vughs and some skew planes, appear at random inth e matrix. Evidence of biological activity since the time of sediment deposition, about 250 years ago, isapparen t from theexistenc e ofchannels ,occurrin g regularly in sizes of up to0. 2 mm, root remains, and pedotubules (aggrotubules and isotubules) up to 10m m in diameter. This already points towette r soil conditions in the past, due to theperennia l flow of spring-water carried by thewad i bed,whe n the stream channel had aggradated to itsmaximu m level, some 4 meter higher than today, during theOttoma n period.

AUTHIGENIC PYRITE Conclusive evidence of such a high level ofth ewad i bed is provided by the presence ofpyrite , which can only form under reduced water-logged condi­ tions. A cluster ofpyrit e framboids (Photo 39), 0.01-0.04m m in diameter, obviously formed in situ, is situated in a channel with root remnants. A weakly developed neoferran (growth of iron-oxides inside the soil matrix)

162 PHOTO 39.Pyrit e framboids (0.03 mm) PHOTO 40.Ferri c nodule (0.3mm )i n the in theuppermos t sedimentary layer uppermost sedimentary layer (incident light). (plain polarized light). adjacent to thepyrit e cluster and thechanne l edge, probably formed as are ­ sult of partial oxidation ofth e pyrite. Other ferric nodules (Photo 40) in the matrix confirm thispictur e ofwe t conditions inth erecen t past. To the trained eye,wea k gley features areeve n visiblemacroscopicall y inth e field. The observed gley phenomena, neo-ferrans and ferric nodules, are usually associated with an alternating oxidation-reduction environment. Such condi­ tions may have occurred at a certain level X: (a)durin g aggradation, just before the stream bed and the water flow arrived at level X, or (b)whe n the stream bed and water flow were at some vertical distance above level X, so that downward seepage ofwate r was not sufficient to fully saturate the soil and exclude the entry ofai r (oxygen); (c)durin g downcutting, as the position of the stream bed moved slightly below level X. A full reduction environment with pyrite formation in the Kadesh-Barnea valley could only occur in those sediments situated atth e actual level of water flow or to some depth below this level, as long as the soil matrix remained saturated with water,excludin g the entry ofair .

163 Some secondary crystallization of calcium carbonateha sclearl y taken place inth e recently depositeduppe ­ rmost alluvial loessial sediments. Apart froma ver y few incipient car­ bonate nodules, some nicecalcitan s occur inchannels , often related to roots.Newl y formed calcitecrystals , 0.002-0.03 mm insiz e (fine silt), completely surround a root remnant having a diameter of0. 3 mm (Photo 41). Even smallermicrit e crystals (0.002-0.01)appea r ina cluster, 0.1 mm indiameter , attached to a root. The occurrenceo f gypsum is not very apparent, but very small,newl y formed crystals (0.005-0.02mm )see m to makeu pa cluste rwit h asiz e of 0.06-0.15mm . Thedegre eo fsoi ldevelopmen t in PHOTO 41.Smal l pedogenetic calcite such ashor tperio d oftim e ofabou t crystalsaroun d aroo t remnant ina 250year s iscertainl y remarkable for deposit lesstha n30 0year sol d (crossed polarizers). a desert region. However,on e should take into account the very large contrasts inenvironmenta l circumstances: First water-logged towe t from the perennial flowo fwate r inth ewad i bed during acertai nperio d after deposi­ tion, when aggradation inth e stream valley reached itshighes t level.Later , probably in the 18thcentury , theproces so fdowncuttin g followed therelati ­ vely long period ofaggradation . Suddenly thesoi lenvironmen tbecam ea sdr y as the surrounding desert, when the stream channel lowered its bed.Onl y some capillary rise ofwate r may havemoistene d thisuppe rpar t of the alluvial loessdeposi t inth ebeginnin g period ofprogressiv e downcutting, apart from the periodic wetting ofoccasiona l heavy rainfall thatpenetrate d into the somewhat deeperpar to fth e soil. No irrigation farming seems tohav e taken place at thisparticula r spot in thevalley . Because of this sudden transition from aquatic to arid conditions, the pyrite hasbee n preserved sowel l in this soil, aswel la s inth emorta r of theaquaduct salread y described previously. Incas eo fprolonge d moistoxida -

164 tion conditions during the transition from wett odry , the pyrite should probably have disappeared already (Van Breemen, 1986,persona l communication).

MICROMORPHOLOGY ANDPALEO-ENVIRONMEN T OF KB-131 The age of thisalluvia l loessial layer isabou t 400 years. The soil matrix is denser than in the uppermost sedimentary layer (KB-129) described above. The related distribution ofplasm a and skeleton grains is,accordingly , mostly porphyroskelic and occasionally agglomeroplasmic. The clay content appears to be higher and there is, consequently, more plasma than skeleton grains. The latter particles are similar incomposition ,siz e and shape as in the previous sample. A few pieces ofcharcoal , up to2 mm ,an d fossiliferous carbonate grains appear randomly in thematrix .Th e largest solid component is a subrounded piece of limestone,2. 2 x4. 6 mm insize . Past biological activity in the soilmus t have been intense,a s indi­ cated by the many channels, root remnants and pedotubules. The latter features are caused by small digging and burrowing creatures that moved through the soil.Th e following types of pedotubules can be distinguished in the sampled layer: granotubules, isotubules and aggrotubules, varying in diameter from0.7-3. 0 mm (Photo 42). Apart from channels, various other typeso fvoids , like vughsan d skew planes occur at random or clus­ tered in the soilmatrix .

AUTHIGENIC PYRITE Conditions of a water-logged reduction environment in thepas t is clearly indicated by the presence of PHOTO 42.Pedotubul e 3m m in pyrite framboids, which regularly diameter (crossed polarizers). occur in thematrix , usually related to voids. The pyrite crystals are often weathered. Evidence ofwe t conditions and alternating oxidation-reduc­ tion circumstances isals o provided by the occurrence ofneo-ferra n cutans,

165 0.02-0.05 mm thick, and small diffuse ferric nodules. Gley features are somewhat more developed and pyrite appears in amor e advanced stage ofweathe ­ ring,a scompare d toth euppe r sedimentary aggradational layer. Some neo-formation ofcalcit e has taken place, usually inth e matrix as very small-scale incipient concentrations, not easily identified as such.On e crystal chamber was observed in a vugh of0. 4 x 0.5 mm, containing fine micritic and sparitic calcite crystals, 0.005-0.05 mm in size. Considerably more widespread in this layer is the crystallization of gypsum in voids. Several channels and vughs are filled up by gypsum crystals, both in needle and more equilateral shapes. Not all the gypsum isnecessaril y derived from pyrite oxidation, as itma y also have crystallized above the level of stream flow from capillary rising water.

MICROMORPHOLOGY ANDPALEO-ENVIRONMEN TO F KB-132 This layer was approximately deposited in 1500A.D. ,jus t above the first step of the terraced Iron Age slope, at the same level as the base of the buttress wall. Themicromorpholog y is in many respects similar to the former sample from a layer 75c m higher. Therefore, only theremarkabl e features or differences with the former sample will be described. A very large striotubule, 2 cm wide and 11c m long, appears in the middle part of the entire thin section. Biolo­ gical activity and pedoturbation has clearly been intense in thepast . The amount of gypsum is markedly higher than in the former sample.Man y channels PHOTO 43.Channel s filled with fine gypsum crystals and other voids are (length photo = 1.5 mm;crosse d polarizers). completely filled in with fine gypsum crystals, 0.005-0.04 mm in size (Photo 43). As some voidswithi n the very large striotubule, mentioned above, are also filled with gypsum crystals,i t

166 can be concluded that the phase of intense biological activity preceded the period ofgypsu m crystallization. Apart from two large, weathered pyrite framboids, 0.08 mm and 0.12 mm in diameter, lesspyrit e seems tooccu r ascompare d toth e samples above. Since this layer isclose r situated to the present water level, moist oxidation conditions appear tohav epersiste d for a longer period oftim ea t this level during the successive phases of aggradation and downcutting. Most of the pyrite has probably disappeared asa result . Thewidesprea d occurrence of gypsum may underline this viewpoint, although most ofth e gypsum has probably been derived from capillary rising water and subsequent crystallization. From the micromorphology samples so far described, itca n be concluded that the transition from awater-logge d reduction environment to the very dry regional desert environment has been particularly sharp,withi n ashor t period of time, in the aquaduct on top of the 7th century dam, upstream from the spring, and inth e uppermost aggradational sedimentary layer in the profile south of the tell. Pyrite,forme d in the latter aquaduct (Photo 27,28 )durin g the time of functioning of the 7th century dam, or formed in the uppermost aggradational deposits (Photo 39), inth e 17th-18th century A.D.,ha s clearly been best preserved because of this sharp transition.

GRANULOMETRY ANDCHEMICA L ANALYSISO F THE AGGRADATIONAL DEPOSITS SOUTH OF THE TELL Eight sampleswer e taken from the alluvial loessial deposits south of the tell (Figure 26) forgranulometri c and chemical analysis. Three of these samples are from the same respective positions as themicromorpholog y samples (KB-129, KB-131, KB-132), already described above,an d have,therefore ,kep t the same respective sample numbers. The following table defines the location of the various samples with respect to the buttress wall, the present wadi bed, the verticalgrid-syste m used during thearchaeologica l excavations by Rudolph Cohen,an d the depth below the former valley plain surface at the time ofmaximu m aggradation. The former valley plain surface near the stream channel wasa t aleve l of about 19.00m during thephas e ofmaximu m aggradation, expressed within the vertical grid-system used during the excavations. The upper surface near the stream channelha sbee nerode d sinceth eonse t ofth eperio d of downcutting and incision. Very recently alaye r of debris from theexcavation s was pushed over the eroded surfaceb y abulldoze r (Figure26) . The particle sizeanalysi s reflects the loessial character of the alluvial

167 TABLE 21. Position and approximate age of samples from the alluvial loessial fill sediments in the Kadesh-Barnea valley, south of the tell, deposited during the period ofstrea m channel aggradation from the 13-18th century A.D.

Sample Depth Height within Distance Approximate time (cm) the vertical South of North of of sedimentation grid-system of buttress present (A.D.) the excavations wall wadi-bed (m) (cm) (cm)

KB-129 25- 50 18.75-18.50 125 1050 17-18th century KB-130 60- 95 18.40-18.05 225 950 17-18th century KB-131 120-140 17.80-17.60 350 825 16-17th century KB-132 190-220 17.10-16.80 300 875 16-17th century KB-136 160-180 17.40-17.20 600 575 16-17th century KB-135 230-250 16.70-16.50 600 575 15-16th century KB-134 280-300 16.20-16.00 600 575 14-15th century KB-133 350-370 15.50-15.30 600 575 13-14th century • 409 14.91 1175 0 Present wadi-bed # 16.94 0 1175

* Water flow in present 20th century wadi-bed. # Base-level of Iron-Age buttresswall . fill deposits, inwhic h silt isth e dominant component. The lower section of the fill is somewhat more clayey than its upper part,althoug h a layer ofcla y loam (KB-131)exist swithi n the upper section aswell . The total amount ofsalt s in the alluvial fill sediments south ofth e tell gradually decreases withdept h (Table 22). The salinity ofth e upper layers can be rated from moderately high (EC8. 4 mS/cm)t omediu m (EC 5.8 mS/cm), further decreasing downward to a low salinity (EC 1.1 mS/cm)i n the deeper layers. Such acharacteristi c distribution evidently shows the salts to be derived from capillary rising spring-water flowing over the wadi-bed, as suggested byKamphors t (1985,persona lcommunication) . Sodium and chloride are the dominant salts inth e fill sediments, followed by sulphate. This seems tob ecompatibl e with thecompositio n of the spring- water (Table 22), from which most of the salts are thought to be derived. Samples KB-136 and KB-135ar e situated near recent erosional surfaces, now covered by a layer ofbulldozer-pushe d debris from the excavations. This proximity to an erosional stream channel surface might perhaps relate to the relatively larger contents ofbicarbonat e anions and sodium-carbonates. The pH is higher as aresult . Themicroscopicall y observed presence ofgypsu m fits with themeasure d sulphate content,bein g highest in sample KB-132 (Photo43) .

168 TABLE 22. Granulometry and chemical analysis of alluvial loessial fill sedimentsi nth eKadesh-Barne a valley, southo fth e tell,deposite d duringth eaggradationa l period from the 13-18th century A.D.

SAMPLE DEPTH CLAY SILT SAND TEXTURE GRAVEL CARBONATES <0.002 0.002- 0.063- >2m m CaCOj mm 0.063 2m m (cm) (%) (%) (55) (») («) KB-129 25- 50 14.4 41.4 44.3 LOAM 2 42.8 KB-130 60- 95 14.7 47.1 37.9 LOAM 0 42.7 KB-131 125-140 28.4 48.3 23.2 CLAYLOA M 0 45.3 KB-132 190-220 15.2 49.1 35.8 LOAM 0 41.9 KB-136 160-180 15.9 50.8 33.3 SILTLOA M 0 45.4 KB-135 230-250 23.6 40.9 35.5 LOAM 0 46.0 KB-134 280-300 25.9 47.3 26.8 LOAM 0 48.1 KB-133 350-370 23.4 45.0 31.5 LOAM 0 46.6

SAMPLE DEPTH pH EC(1:1) SAR ORGANIC TOTAL rNITROGE N CARBON PHOSPHORUS (cm) (mS/cm ) (K)C (S)N (8) P2 °5 KB-129 25- 50 7.8 8.40 52.9 0.b4 0.16 0.03 KB-130 60- 95 7.1 6.58 31.4 0.27 0.14 n.d. KB-131 125-140 7.4 5.98 48.6 0.81 0.18 n.d. KB-132 190-220 7.4 5.78 47.9 0.84 0.17 n.d. KB-136 160-180 7.8 3.62 54.1 0.27 0.15 n.d. KB-135 230-250 8.1 1.10 30.0 0.83 0.17 n.d. KB-134 280-300 7.7 1.08 30.3 0.88 0.18 n.d. KB-133 350-370 7.7 1.85 33.3 0.43 0.15 n.d.

SAMPLE DEPTH SOLUBLESALT S IN 1:1SOIL/WATE REXTRAC T (meq/kgsoil )

2+ (cm) Cra Mg Na+ K+ CO3"HCO 3 Cl" SO4" NOJ

KB-129 25- 50 3.5 <0.10 70 .0 0 0 3 7 56.0 12 0 5.0 KB-130 60- 95 3.8 <0.10 43 .5 0 0 1 2 33.0 14 0 0.6 KB-131 125-140 1.6 <0.10 43 .5 0 0 1 5 34.0 11 0 0.6 KB-132 190-220 1.6 <0.10 43 .5 0 0 0 9 30.5 17 5 0.2 KB-136 160-180 1.2 <0.10 42 .8 0 0.2 2 2 30.5 9 0 1.8 KB-135 230-250 0.3 <0.01 10 .6 0 0.2 3 6 5.8 1 1 0.4 KB-134 280-300 0.3 <0.01 10 .7 0 0 2 8 6.7 1 8 0.2 KB-133 350-370 0.6 <0.10 19 .0 0 0 2 1 14.0 3 6 <0.1 SPRING-WATER40 5 4.4 3.7 14 .4 0.1 n.d.3 5 14.4 4 7 n.d.

Percentagesi nweigh t %o fth e fine earth< 2mm . Only graveli nweigh t %o f the total soil. EC= electrica l conductivity, SAR= sodiu m adsorption ratio. n.d.= no tdetermined .

169 AN EXCAVATED SOIL PITWES T OF THE TELL The pit was excavated to determine the stratigraphy of the upper sedi­ ments at this part ofth e valley, aswel l as theparticl e size distribution and main soil chemistry. The location of the pit is3 8m wes t of the tell and some 60m north ofth e stream channel.Fro m an agricultural point of view, the pit lies next to a field with olive trees (Photo 37), farmed by the Bedouin. The surface of the pit is situated at a level of 18.29m ,expresse d within the vertical grid-system used during the archaeological excavations. The sedimentary stratigraphy until the depth ofexcavatio n (1.50 m) is essentially twofold:

1) 0-75 cm Fine loessial deposits with a texture of loam, containing some scattered stones. 2) 75- 150+ cm Gravel with stones up to 10c m insize .

Another excavated soil pit, located at a distance of about 10m eas t from the north-eastern corner ofth e tell, exhibits a similar sedimentary strati­ graphy. The boundary between the loessial sediments and the underlying gravel liesher e at a depth of6 0cm . The surface of thispi t is situated at a level of 18.61m , expressed within the vertical grid-system of the archaeological excavations. The resulting stratigraphic picture of this part ofth e valley, aseluci ­ dated by the investigated pits and channel sections,clearl y shows the blanket of alluvial loessial sediments to increase in thickness toward the stream channel,whils t the loam-gravel boundary descends accordingly. The particle sizedistributio n of the upper two horizons in the soil pit west of the tell israthe r similar to the granulometry of the upper fill sediments adjacent to the stream channel, south of the tell (Table 22, 23). This might indicate that some alluvial loessial sediment was also deposited farther away from the stream channel,durin g theperio d ofmaxima l aggradation in the 17-18th centuries A.D. Sediment laden floodsprobabl y covered the entire Kadesh-Barnea valley when channel aggradation reached its highest levels inhistorica l times. Even in the beginning of the 20th century large runoff floods could occa­ sionally sweep the whole valley, as reported by Woolley and Lawrence (1936:77). The marked difference being, that the flood reported by the British authors caused erosion, whereas the floods during the period of aggradation had the opposite effect,causin g sedimentation!

170 TABLE 23 Granulometry an dchemica l analysis ofa nexcavate d soilpi ti nth e Kadesh-Barnea valley, next to an irrigated field with olive trees, 38m west of the Tell.

HORIZON DEPTH CL.AY SILT SAND TEXTURE GRAVEL CARBONATES <0.002 0.002- 0.063- >2 mm CaC03 mm 0.063 2 mm (cm) (») (S) («) {%) («)

A 0-15 12.5 48.1 39.4 LOAM 1 45.0 C-I 25-35 14.0 45.8 40.2 LOAM 2 45.5 C-I 45-55 19.8 31.3 48.9 LOAM 8 46.4 C-II 70-80 20.0 26.1 53.8 SANDY-LOAM/GRAVEL 24 49.4

HORIZON DEPTH pH EC(1:1) SAR ORGANIC TOTAL CARBON PHOSPHORUS (cm) (mS/cm) (K)C . {%) P2O5

A 0-15 8.3 0.63 15.8 0.69 0.11 C-I 25-35 8.2 1.22 39.1 0.69 0.16 C-I 45-55 7.9 1.25 43.0 0.71 0.14 C-II 70-80 7.2 7.00 51.3 0.80 0.14

HORIZON DEPTH SOLUBLE SALTS IN 1:1SOIL/WATE REXTRAC T (meq/kgsoil )

2 (cm) Ca2+ MQ2+ Ne + K+ CO2" HCO" CI" SO " NQ- 3 3 4 3 A 0-15 0.5 <0. 01 7. 9 <0.10 0 5.4 3 0 0.5 <0.1 C-I 25-35 0.2 <0. 01 10. 7 <0.01 0.8 0.1 5 2 4.6 <0.1 C-I 45-55 0.2 <0. 01 13. 6 <0.01 0 2.5 6 2 5.0 0.1 C-II 70-80 5.7 <0. 10 87. 0 <0.10 0 1.0 49 0 38.5 0

Percentages inweigh t %o fth efin e earth <2mm . Only gravel inweigh t %o f the total soil. EC= electrica l conductivity. SAR= sodiu m adsorption ratio.

The amounto fsolubl e saltsi nth euppe r6 0c mo fth esoi l profileca n be rated very lowt olow , increasing slowly with depth, untilth esal t content showsa ver y sharp increasea tth eloam-grave l boundary, ata dept ho f7 0cm . Such a salt distribution isi nmarke d contrast totha to fth e fill profile south ofth etel l (Table 22), andreflect sth einfluenc eo fleaching , duet o irrigation farmingb yth eBedoui n onth eadjacen t field. Sodium isth edominan t cation throughoutth eprofile ,wherea sth eamoun to f bicarbonate surpassesth echlorid e anion content inth eA horizon . Theamoun t of sulphate anionsi nth ethre e subsoil horizonsi snearl y aslarg e as the chloride anion content, which isals o understood tob ea resul to f leaching with irrigation water fromth espring . Thenitrat e content is consistently lower than inth efil lprofil e southo fth etell .

171 STREAM CHANNEL EVOLUTION INTH EKADESH-BARNE A VALLEY DURINGHISTORICA L TIMES

In conclusion ofth e detailed descriptions about the stratigraphy, radiocarbon dating, mortar micromorphology and soil development of various investigated sections inth e valley ofEi n el Qudeirat, the most relevant stratigraphic information isbrough t together in Table 24. On the basis of these data, an attempt ismad e to reconstruct the evolution of the stream channel since theLat eBronz eAge .

TABLE 24. Geomorphic, archaeoloqic, and C-14 data about stream channel evolution in the Kadesh-Barnea valley during historical times, in chrono-stratigraphic sequence.

Object Below Above C-14 Conventional Calibrated present present charcoal C-14 date C-14 date valley wadi- sample (BP) (historic plain bed years) (cm) (cm)

Surface aquaduct 230-310 215-295* GrN-12327 3270 + 100 1725-1425 BC Iron Age wadi-bed >400 <0 Basis dam 257-400 73-203* GrN-12326 1380 + 90 610- 685 AD Wadi-bed aquaduct 410-510 15-115* GrN-12328 1395 + 50 615- 665 AD Charcoal lense 300 7 QC-491 665 + 115 1260-1400 AD Charcoal pocket 250 150 GrN-11946 540 + 45 1310-1430 AD Charcoal pocket 300 130 GrN-11947 445 + 25 1440-1460 AD Charcoal lense 250 150 GrN-12333 410 + 35 1450-1480 AD idem idem idem GrN-11949 395 + 45 1450-1485 AD idem idem idem GrN-12331 340 + 40 1470-1530 AD Charcoal lense 30 375 GrN-12332 70 + 170 1665-1950 AD

# The range in heights expresses the vertical thickness of the aquaduct. * The range inheight s expresses maximum and minimum levels of present wadi bed and valley plain levels.

During theLat e Bronze and Iron Age, the surface ofth e valley plain was apparently somewhat lower than today,particularl y in its narrow part upstream of theJarvi sda m and inproximit y toth e stream channel. The level of the stream bed appears also tohav e been somewhat lower than at present. This status quo or dynamic equilibrium inth e stream channel of the valley of Ein

172 elQudeira t may have continued, albeit with presumed small fluctuations, since the Iron Age probably until Late Hellenistic or Roman times. During theLat e Byzantine period the stream bed level was about one meter higher than today, taking into consideration thebasis-leve l of the 7th century (A.D.) dam, as well asth e level ofth e 7th century wadi-bottom aquaduct that tapped thewate r flowing over the stream bed. Since the wadi bed seems to have been below the present level during the Iron Age, stream channel aggradation is likely tohav e occurred at some time before the Late Byzantine Period. Acharcoa l sample collected by Goldberg (198A) from a fill deposit 3k m northwest ofQuseima ,abou t 9k m northwest of Ein elQudeirat , seems to fit this picture.Th e sample (QC-492)yielde d aC - 14 date of 1755+ 105B P in conventional carbon-14 years. When calibrated against the high precision calibration curve ofPearso n et al. (1985), the resulting outcome (Figure 29)in historical years is 125-405Ca lA.D .

1860 -

1800•

1600

125 200 240 300 405 500 HISTORICAL YEARS A.D.

FIGURE 29.Calibratio n ofth e above C-14dat e (QC-492)int o historical years.

The destruction ofth e 7th century dam coincided perhapswit h aperio d of stream channel incision during the Early Arab period, by which the stream bed was lowered with about 1-1.50m belo w the Late Byzantine level. This can be concluded from the sedimentary stratigraphy near the 7th century dam.

173 Themos t extraordinary period of stream channel aggradation during histori­ cal times apparently began in the 13th century A.D.,durin g the latter part of theCrusade r period or very beginning of theMamlu k period, and lasted until about the 18th century. The stream bed rose with about 4mete r during this aggradational period! This remarkable regime of aggradation inNaha lKadesh-Barne a came to an end apparently in the 18thcentury . It was followed by adramati c change that led to rapid downcutting and incision during the last fewcenturies ,whic h lowered the stream bed with about 4mete r to itspresen t level. This process of incision and erosion seems to continue even today.

TABLE 25.Summar y conclusions of stream channel evolution in the Kadesh-Barnea valley during historical times.*

HISTORIC PERIOD APPROXIMATE TIMING STREAM CHANNEL BEHAVIOUR

Late Bronze Aget o 1500- 100B.C . Dynamic equilibrium ofth e Late Hellinistic- stream channel,wit h small Roman times fluctuations.

Late Hellenistic 100B.C . -40 0A.D . Stream channel aggradation - Roman Period and dynamic equilibrium

Byzantine Period and 400- 700A.D . Dynamic equilibrium Early Arab Period

Early Arab Period to 700-120 0A.D . Incision and Late Crusader Period dynamic equilibrium

Late Crusader Period 1200-170 0A.D . Major aggradational period toLat e Ottoman Period

Late Ottoman Period 1700-198 6A.D . Rapid incision and erosion until the present

* This summary framework isbase d upon available evidence in the present state ofknowledg e and understanding. New data may refine or alter thispicture .

174 11 Water, food production and population in the Kadesh-Barnea area

INTRODUCTION

There are usually several criteriat oasses s past population levels ina certai n area. Thenumbe ro fbuildin g structuresi scommonl y usedi nthi s respect (Broshi, 1980). However,i ncase so ftent-dwellin g communities,lik e the ancient Israelitesdurin g their sojournmenti nth e desert after the Exodus fontEgypt ,o rothe r nomadican dsemi-nomadi c inhabitantso fari d regions,th e latter criteriumi sgenerall yo fn opractica l value.Th epitchin go ftent san d their subsequent removal will hardly leavean ytrace , ifa t all.I n these circumstancesa nasessmen to fth e natural food producing capacityo fa certai n dry region, incombinatio n with the availabilityo fwater , mayb eth e only helpful criteriuma sa paleo-populatio n indicator. With regardt oth eKadesh-Barne a valley, iti spossibl et omak ea nestima ­ tion of food production levels based upon irrigation agriculture with water from the springo fEi ne lQudeirat . Sucha napproac hi si na sens e hypothetic and tentative. Itseems , nevertheless, very usefult oasses sth e carrying capacity ofth e areai nterm so ffoo d production. Thenumbe ro fpeopl e that can befe d maximallyi nth eKadesh-Barne a valley, exclusively relying on a self-sufficient typeo fsubsistenc e farming, doesno tnecessaril y coincide, however,wit h actual population numbersi ncertai n historicperiods . The spring wascertainl yno tuse dt oit sful l potentialfo rfoo d production inmos t periodsi nth epast ,a sthi s would require fullcontro l over the water flow from the spring, night storage,an dtw o food cropspe r year,bot hi nth e winter andth e summer season. Itseems , therefore, likely that population levels directly supportedb yfoo d produced with water from the spring were

175 usually below themaximu m potential. On the other hand, population numbers might have been well above themaximu m carrying capacity ofth e valley, as food could have been supplied through imports or by othermeans . The major food crops, both inantiquit y and atpresent , are thecereals . About 75 %o fou r food today is derived, one way or theother , from cereals (Buringh, 1977). Leguminous pulses or beans (Westphal, 1974,1982 )ar e second in importance, both for their significant protein content and for their impor­ tance in crop rotation tomaintai n acertai n level ofsoi l fertility. Apart from the two-field system (one year a field issown ,th e next year it is left fallow), used in theMediterranea n area (Slicher van Bath, 1960:68), the so-called three-field system was also known inth eregio n in antiquity, especially since itspromotio n inRoma n times. Inthi s system one part of the land was sown with cereals, a second with legumes and the third was left fallow. This fallow land was used for pasturage and became thus fertilized in a natural way by animal manure. The next year thisdivisio n was rotated, by which legumes, usually lentils, were sown in the fallow land of the former year; wheat or barley was sown in thepar t occupied the year before by legumes; and the land inwhic h cereals had grown was left fallow. Thistradi ­ tional system oflan d use and crop rotation persisted in the Southern Levant for many centuries intomoder n times (Turkowski, 1969). Food crops likewheat , barley, lentils and chick peas are known in the region at large from antiquity (Zohary, 1973; Zohary and Hopf, 1973; Hopf, 1978). Most ofthes e food crops grow in thewinte r and spring season, and would require some 300m m ofwater , well distributed during their lifecycle . The average annual rainfall at present isabou t 75m m , usually concentrated in thewinte r months. It seems that summer cereals like sorghum were mainly introduced in theSouther n Levant by the Arabs (Watson, 1983), at some time after the Muslim-Arab conquest in the 7th century A.D. Summer crops would require more irrigation water than winter crops, due toth e higher tempera­ tures in summer and the total lack ofrainfall . Someover-irrigatio n beyond the actual need ofth ecro p is required, due to the slight salt content in the spring water ofEi n elQudeira t (1.42 g/1).

ESTIMATED MAXIMAL FOOD PRODUCING CAPACITY ANDRELATE D POPULATION LEVELSO F THEKADESH-BARNE A VALLEY IN ANTIQUITY AND TODAY.

The purpose of the following hypothetical calculations is to assess

176 themaxima l possible carrying capacity of the area inhistorica l times, rela­ ting food production levels to population numbers, based on a rather autarkic food economy. Rainfall amounts during historical timeshav e naturally varied, as will be discussed in the last chapter, but it seems unlikely that the average annual amounts deviated with more than about 50 %fro m the present average quantity. Thedischarg e of the spring may, likewise, have varied in the past, but these variations are probably alsowithi n a 50 %rang e from the present values. The 20th century precipitation and discharge data served asa basis for the calculations. Such oversimplifications are inevitable in this kind of assessments, by which care is taken not tounderestimat e the carrying capacity of the area. The outcome of the calculations can, therefore, be regarded as a likely maximum indeed. The spring has an average discharge of about 60cubi c meter ofwate r per hour. This amounts to525.60 0cubi c meter ofwate r peryear . Ifa rotational food crop pattern could bearrange d that would allow for an evenly distributed consumptive water use ofth e spring water throughout theyea r (without summer cereals or summer pulses this is actually aproblem) , then the 525.600 cubic meters of annual spring water might be distributed overa n area of 131.4 hectare, by which each field would receive 400mm/yea r as an average. Full control of the spring flow and a system ofeithe r night irrigation or night storage is imperative to use the total yearly discharge ofth e spring for irrigation agriculture. The Jarvis irrigation scheme in the area, already described, which concen­ trated on cash crops, mainly olives trees, brought some 120h a into cultiva­ tion (Jarvis, 1938:267). This shows that enough potentially arable land exists in the area in relation to the capacity of the spring. Hence soil isno t the limiting factor in theKadesh-Barne a valley and the adjacent valleys down­ stream. In antiquity, a yield of 1000kg/hectar e ofcereal so r pulses was consi­ dered to be a good figure, whichcoul d only have been obtained through proper crop rotation and maintenance ofsoi l fertility. The average yieldso f wheat and winter legumes amounted toonl y 530kg/h a on a rainfed fellah's farm in the Yizreel Valley (483mm/yea r on the average), during the second and third decades of the 20th century, in a non-optimal autarkic agro-ecosystem (Stanhiil, 1978). The Nizzana papyri of the6t h and 7th century A.D. mention the yield ofwheat , grown in a runoff farming system in the Nizzana area during acertai n year,t o be 6.9 times the amount sown (Kraemer, 1958). Inth e chapter on HorvatHaluqim , it was already extensively discussed that average

177 wheat yields under runoff farming conditions inantiquit y probably did not exceed 600kg/ha . With proper irrigation farming inth eKadesh-Barne a valley, and a secured supply of about 400m m ofwate r per growing season (abouthal f a year), a yield of 1000kg/h ao fcereal s and leguminous pulsesmigh t have been possible in the past. If food crop production and agricultural water use from the spring of Ein elQudeira tcoul d havebee ndistribute d evenly throughout the year (winter and summer crop), then about 130,000k go fcereal s and legumes might have been harvested from some 130hectares . This quantity hast ob ereduce d by the seed stock required for next year (about 13,000kg )an dpost-harves t losses (Hall, 1970; Vande rLee ,1978 )o f some 10 %(13,00 0 kg), which leavesa nestimate d 100,000kg/yea r for human consumption. Therathe roptimisti c yield ratioo f 1:10 underlines thepurpos e of the calculation toasses sth emaximu m carrying capacity of the area. A detailed study ofhistori c yield ratios in severalEuropea n countries wasmad e by Slicher vanBat h (1963a,b). The averagenutritiona l value ofcereal s and pulses isabou t 3000-3500cal/k g (Westphal,1974 ,1982 ;Ilaco , 1981). Areaso ­ nably well-fed humanbein g needssom e 2500cal/da y or912,50 0cal/year . Ifal l his/her energy requirements were tob e supplied by the 100,000k g of cereals and pulses, 3250x 100,000= 32 5millio n food calorieswoul d beavailabl e per year. The number ofhuma n beings that could liveo nthi s amount would be 325,000,000 : 912,500 =35 6peopl e (about28 0k go fcereal san d pulses per person/year). Assuming alowe rconsumptio n of200 0cal/person/day , which was approxi­ mately the average for India in 1958 (Weitzan d Rokach, 1968), 445 people could have lived onthi samoun t (225k g ofcereal san dpulse s foron e person per year). These figureso f28 0an d 225kg/yea r percapit a arecomparabl e with the figureo f25 0k ggrain/person/year , takenb yBrosh i (1980)a sa n overall minimum average. However, apart from food calories, also theamoun t of food protein hast ob etake n intoconsideration , aswel l asth enee d for vitamins and minerals from fruitsan d vegetables. Van der Woude (1963) studied the consumption ofcereals , meatan d butter, inrelatio n tocalor y requirements, inHollan d at theen d ofth e 18th century. The average daily protein requirement fora huma nbein g isabou t 80 g/day or 29.2kg/year . Aroug hestimat e ofth e protein content in the mixture of 100,000k go fcereal san d pulseswoul d be about 12 %, which amounts to 12,000 kg protein. This quantity seems sufficient forth e yearly requirements of about 411 people, showing that quantitatively enough protein would bepresen t

178 in the cereals and pulses for about 400 people. However, qualitatively, cer­ tain amino acids necessary for thehuma n body areusuall y lacking in a diet entirely composed ofcereal s and pulses. With a limited choice of food crops, soya bean was not known in the area in antiquity, some2 5% of the daily protein intake ought, therefore, to be animal protein (Weitz and Rokach, 1968). Sheep and goats were certainly kept in the area inth epast , grazing the stubble of harvested field crops, fallow land, and the surrounding country­ side. Animal manure would have been important tokee p soil fertility and cereals/pulses production on a level ofon e ton per hectare. The relation between agriculture and pastoralism during historical timesmigh t have varied in between the semi-nomadic pastoralism of theBedouin , to forms of herdsman husbandry and sedentary animal husbandry, as defined byKhazano v (1985). In order to feed the herds inth e difficult months from the second part of the summer until the early winter, in which there isvirtuall y no natural pasture (Hillel, 1982:196), some of the irrigation water islikel y tohav e been used for the growing of forage crops. A certain amount ofvegetable s and fruitswer e probably also grown to enlarge the dietary composition, thereby satisfying taste and nutritional requirements. The food basket was perhaps for 75-90 %compose d ofcereal s and pulses, the remainder made up by animal products, fruits and vegetables. It is, therefore, unlikely that all the available irrigation water,durin g cer­ tain periods in thepast , was entirely used for thegrowin g of cereals and pulses. The latter production was, therefore, probably somewhat lower than calculated above. Moreover, a certain percentage of the food grains might necessarily have been given to the flocks assupplementar y feed. It thus seems likely that with amor e varied dietary composition, population levels would have been somewhat lower, than the number of40 0calculate d on the basis of cereals and pulses alone. However,th e grazing ofherd soutsid e the valley in winter and spring might have compensated somewhat forth e loss in cereal production, because part of the naturally produced biomasso fa n adjacent area isharveste d through theanimals . The maximum possible population level fed by the varied produce of irriga­ tion farming in the Kadesh-Barnea valley inantiquity , including the keeping of some flocks, can thus beestimate d at less than 400people . Thiscalcula ­ tion isbase d on the ideal situation of two harvests of food crops per year,a yield of one ton/hectare and a fully effective useo fal l the water produced by the spring,supplyin g an average amount of 400m m toeac h crop.

179 Inmoder n times,assumin g free access to the necessary chemical fertilizers and an ideal fertilizer-water management,yield s of 8000kg/h a may be possible in the region with the same quantity of about 400m m ofwate r (DeWit , 1958; Van Keulen, 1979, 1980). Thus the "absolute"maximu m food producing capacity of theKadesh-Barne a valley might be sufficient to feed about 3000people . Concerning the direct use ofwate r by man, for drinking, washing, and cooking, the amounts required are relatively low fora population of 400 people. Personal water requirements are estimated at 10lite r per capita per day as aminimu m (Evenari et al., 1971:148-150). For 400peopl e this amounts to 1460cubi c m ofwater/year ,whic h isproduce d by the spring in 24hours .

COMPARISON OF ESTIMATED FOOD PRODUCTION AND RELATED POPULATION LEVELSWIT H ARCHAEOLOGIC DATA OF LATE BYZANTINE -EARL Y ARAB TIMES

It seemshighl y unlikely that full control ofth e spring's annual discharge for agricultural food production was actually realised in antiquity. Perhaps a 75 %efficienc y might have been attained, which would have reduced the calculated maximum carrying capacity from about 400t o 300people . In case food crops, like cereals and pulses, were only grown inwinter , due to the absence of suitable summer varieties inth e region inth e past, then the carrying capacity would have been reduced further from about 300 to perhaps 150 people. The formerByzantin e village in the valley ofEi n el Qudeirat apparently consisted ofabou t 20 houses (Woolley and Lawrence, 1936:80). If each house had 6 inhabitants, thepopulatio n might have numbered some 120 people,whic h isno t far off the latter estimated figure of150 . From the remnants of irrigation systems, attributed to the Byzantine period, Woolley and Lawrence (1936:79,80)conclude d thatmos t ofth e valley of Ein el Qudeirat and beyond wascultivate d inthos e times. The Early Arab period ought to beinclude d inthi speriod , as indicated by the carbon-14 datespublishe d inthi sthesis . The British authors suggested that the local population was in those times probably more numerous than the Bedouins in 1914, who cultivated only part of the valley. Their statement "One can well imagine that inyear s of drought Ain el Guderat fed all the saints of Central Sinai" might indicate, however, an exaggerated expectation of the food produ­ cing potential ofth e valley and its spring. Ifth e latter estimated figure of 150i s somewhere near thetruth , as well as the estimated population for the local village, only 30peopl e could have been fed inadditio n to the villa-

180 gers. Nevertheless, itwa s certainly the only reliable food producing area in the region during drought years. In the early part of the 20th century, night storage was not part of the irrigation system anymore, and fruit trees and summer vegetables were absent, apparently because ofth emalari a hazard, as reported by Woolley and Lawrence (1914/15) and by Jarvis (1938). Insuc h a situation only a quarter of the spring's annual discharge could possibly have been used, being 131,400 cubic meter ofwater . If5 0 %o f this water (65,700cubi c meters)wa s used effec­ tively for winter cereals by the local Bedouin, some 16hectar e could have been cultivated with winter cereals,whereb y each plotwoul d receive 400m m of irrigation water. Assuming a yield of 700kg/ha , some 11000k g of winter grains might have beenproduced . When this amount isreduce d with 40 % for next year's seed stock, animal feed and post-harvest losses, approximately 6600k g would have remained to feed about 33people , ifth e grain composed 80 %o f their diet.

REMAINSO F RUNOFF FARMING INTH E VICINITY OF KADESH-BARNEA

From theprecedin g part it isclea r that agricultural food produc­ tion through irrigation farming with the relatively copious spring ofEi n el Qudeirat was only sufficient to feed a rather limited number ofpeople : from a few tens to a few hundred people at themost , depending upon the degree of water control and farming efficiency. It is,therefore ,no t surprising to find other agricultural systems based on runoff farming inth e immediate vicinity of theKadesh-Barne a valley. Incertai n periods in antiquity, there wasappa ­ rently both an incentive and the capability todevelo p thetw o farming systems side by side in the same region. Thepresen t Bedouin population also uses both systems for farming inthi spar t of the desert,albei t ina low key manner.

THE PLATEAU SOUTH OF THE KADESH-BARNEA VALLEY A number ofterrace d wadisappea r on theundulatin g plateau to the south­ east of the spring ofEi n el Qudeirat. From the remainso fa n apparent four room ancient farm house (coordinates 09695-00675)on e overlooks a terraced wadi lying to the south of it. The terraced wadi, a tributary of Wadi Um Hashim, contains about 20check-dam s spaced at regular intervals across its stream bed. Only the two lowermost terraced fields havebee n ploughed by the Bedouin.

181 GEBEL EL QUDEIRAT,NORT H OF THEKADESH-BARNE A VALLEY A small but neatly terraced wadi appears on themiddl e slopes ofGebe l el Qudeirat, above the Jarvis Dam. This runoff farming system consists of 5 check-dams, the latter dam being almost 2 meter high (Photo 44). The check- damsar e orderly built ofnaturall y shaped stones of approximate similar size. There isn o clear evidence as to the age of the system,which , judged from the amount of silt accumulated behind the terrace dams,mus t date back toantiqui ­ ty. The absence ofnatura l vegetation inside the terraces and the appearance of plough-marks, shows that the system isuse d for food crops by the local Bedouin. The position ofthi s runoff farm on the slopes above the valley of Ein elQudeirat ,s o close to the spring and itsperennia l stream, demonstrates the determination incertai n periods of antiquity to develop and use every possible part ofth e desert for agricultural purposes.

PHOTO 44.A n ancient runoff farm onGebe l elQudeirat ,abov e the Jarvis Dam.

GEBEL EL EIN Continuing from Gebel elQudeira t to Gebel elEin , in the direction of Sheluhat Kadesh-Barnea, a very isolated runoff system appears on the upper south-facing slopes ofthi s mountainous ridge. The system, composed of only four check-dams, is situated in a slightly flat and concave part of the

182 hillslope. The runoff farming system isno t used atpresen t by the Bedouin. The following table shows the measures of damsan d fields.

TABLE 26. Runoff farming system on Gebel el Ein near Sheluhat Kadesh-Barnea, hauinq acultiv/abl e area of only 675 squaremeter .

LENGTH WIDTH HEIGHT FIELD WIDTH FIELD AREA (m) (m) (m) (m) (m2)

DAM 1 30.80 0.60 0.30 6 185 DAM 2 22.00 0.70 0.50 8 210 DAM 3 17.30 0.70 0.30 7 140 DAM 4 16.20 0.90 0.65 8. 50 140

WADI ELHALUF I Quite a large number of runoff farming systems occur between Sheluhat Kadesh Barnea and Ein elQudeira t along Wadi elHalufi , atributar y ofNaha l Kadesh-Barnea. These systems are partly eroded, because of incision. The Bedouin use some of the remaining fields for wheat orbarle y cultivation.The y generally dono t repair the ancient runoff systems,which ,a sa resul t receive only direct rainfall, since runoff water cannot reach these fields any more. In this way abarle y crop may only be a successful ina noccasiona l very wet year, perhapsonc e in fiveyears ,a snote d by Jarvis (1938:245).O n the other hand, there are alsoexample s ofBedouin s whohav e repaired or constructed runoff catching dams.

THE WESTERN PART OF GIVAT BARNEA This area, a plateau some 5k m to the east of Kadesh-Barnea, at an average altitude ofabou t 550m , isdissecte d by tributary wadis of Nahal Kadesh-Barnea, many ofwhic h were terraced forrunof f farming in antiquity. The mountainous ridgeo fSheluha t Kadesh-Barnea boundsth eare a to thenorth . A very nice terraced wadi, not eroded, is fully used by the local Bedouins, who have ploughed the fields. The author visited the area on February 21, 1982, when thecerea l crop sown was just beginning toemerg e from the soil. This particular wadi wasprobabl y terraced inth eByzantin e period. From the remains of a Byzantine farm-house oneha s aclea r view over the terraced fields (Photo 45). Itsdam swer e built with large stones in distinct rows. The first dam in focus is 1.80 high and has awidt h of 2m . The average size of itsbuildin g stones is40-5 0cm , but huge blocks of 1.20 mals o occur. This first dam ispartl y built upon bedrock.

183 1 * r-»^S!;%3H?W

PHOTO 45.A tributar y wadio fNaha lKadesh-Barnea , probably terraced inth e Byzantine period, farmed todayb yloca lBedoui n forfoo d crops.

A beautiful circular threshing floor (Photo 46)wit ha diamete ro f19.5 0m is situated southo fth e former wadio na somewha t higher parto fth e plateau. The threshing floori suse db yth e Bedouin,a sindicate db yth e adjacent heaps of ensilaged grain stocks, covered with earthan dbranche s (Photo 47). On a nearby fairly level plateau are the remnantso fa nancien t fortress, another circular threshing floor (19.70m diameter), andancien t circular silo pits (1.30 mdiameter) . Oneo fthes e pitswa s excavatedan dappeare d tob e lined with stones toa dept ho f5 0c mbelo w the surface. Large upright standing stonesa tth e edgeo fth ecircula r threshing floors,ma y have serveda sa kin d of fencei nth epast . From the remnantso frunof f farming systems,threshin g floorsan dsil o pits in the vicinityo fKadesh-Barnea , itseem sclea r that cereal food crops were growni nth e areai nth e past with runoff rainwater, onlya fe w kilometerst o the east ofth e valley oasiso fEi ne lQudeira t with its springan d totally different systemo firrigatio n farming. Both systemso fdeser t farming are also used todayb yth eBedouin .

184 PHOTO 46.Circula r threshing flooran densilage d grain mounda tth eright-han d side. Themountainou s ridgeo fShluha t Kadesh-Barnea risesi nth ebackground .

^2 "°.

PHOTO 47.Ensilage d cereal stocks,produc eo floca l Bedouin runoff farming.

185 THE POST-EXODUS SOJOURNMENT OF THEANCIEN T ISRAELITES IN THE DESERT

Finally, someremark s about thematte r ofencampment ,wate r and food supply of the ancient Israelites after theExodu s and their 40 years in the desert. The areao fEi n elQudeira t and Ein Quseima isgenerall y accepted as themos t likely region inwhic h Kadesh-Barneawa sprobabl y situated. It seemsunlikely , though,tha t the actual encampment of the Israelites was in the valley plain ofEi n el Qudeirat, because of thedange r of flooding in the rainy season. Thewhol e valley was swept by a flood in 1911/12, asmen ­ tioned by woolley and Lawrence (1936:78). During the recent excavations of the tell by Rudolph Cohen (1976-1982), runoff water was occasionally flowing through the tents standing on the valley plain near the tell. The excavated fortresses (Cohen, 1981, 1983) were not built on the general level of the valley plain, but on a somewhat elevated natural hillock. The Byzantine village and theBedoui n encampment are also situated onhighe r grounds,o n the footslope ofGebe le lQudeirat . As already estimated, the valley ofEi n el Qudeirat might have produced in antiquity food forabou t 400peopl e at themost . Ifth enearb y springs of Quseima and ofEi n Qadeis were also fully used for food production, perhaps enough food might have been grown for 700people . If runoff farming was practised in the surrounding area during theLat e Bronze Age, maybe enough food forabou t 1000peopl e and their livestock might have been grown in years without drought.Thi s number is,however , very remote from the large number of people mentioned in theTora h (Exodus 12:37; Numbers 11:21). A food producing economy based on extensive livestock rearing seemsou t of thequestion , as these Israelites were, according to theSciptures ,no t dis­ persed throughout theSina i peninsula, but were living together in the desert in anorganize d manner. Moreover, the widespread assumption that pastoral nomads can or could survive in a self-sufficient autarkic way is a myth, masterly refuted by Khazanov, who states "pastoral specialization has meant more or less economic one-sidedness and no autarky...nomad s could never exist on their own without the outside world and itsnon-nomadi c societies... the important phenomenon of nomadism really consists in its indissoluble and necessary connection with the outside world" (Khazanov, 1985:3). It seems likewise out of the question that the Israelites, sojourning in the desert, could depend for their food supply on the non-nomadic outside world, on Edom, Moab, or on theCanaanit e inhabitants of thepromise d land, which they intended toconquer . Ifthe y had to feed themselves by agricultural

186 and pastoral food production in the region ofQuseim aan d Ein el Qudeirat, their numbers could not possibly have exceeded amer e 1000peopl e at themost . However, the surrounding enemy nations would not have feared such a small group ofpeople ,advancin g toward theirborders . Although the high population numbers mentioned in the Bible are often discounted asmer e symbolic or typological expressions (Broshi, 1980), the context in which these numbers areplace d cannot easily be dismissed. The Biblical narrative doesno t portray the ancient Israelites,havin g come out of Egypt, as a small band of some 1000Bedouins . The Torah clearly exhibits another picture: "And Moab was sore afraid of thepeople , because they were many...". Balak, the king ofMoa bwen t toBalaam , saying: "Behold, there isa people come out from Egypt:behold , they cover the face ofth e earth,an d they abide over against me...the y are too mighty forme " (Numbers 22:3-6). Apart from food supply,ther e isth e problem ofwater . Ifvirtuall y all the waters ofth e springs in the region were fully used for agricultural produc­ tion, there would havebee n nodrinkin g water left fors o large a community. In terms of drinking water the area ofEi n el Qudeirat and Ein Quseima might have sustained a significant population. However,i n termso f food supply,b y which it isunlikel y that the surrounding enemy nationswoul d have exported food to the advancing Israelites, there seems indeed noalternativ e but the unique food provision described in theTorah :manna . The practical problem how so large a community could have been sustained and fed for some4 0year s in the desert, or even foron e year, cannot be simply removed by scepticism and scorn for the subject as such. The matter touches upon the very existentialism of Israel's history. With regard to the demand for meat alone, theproble m iswel l expressed by Moses himself: "...thou hast said, 'Iwil l give them meat,tha t they may eat awhol e month!' Shall flocks and herdsb e slaughtered for them, to suffice them?O r shall all the fish ofth e seab egathere d together for them, to suffice them?" (Numbers 11:21-22). On the one hand,th e region ofKadesh-Barne a and Quseima,possibl y the most favourable agricultural area in the entire Sinai-Negev desert, could not have fed more than 1000peopl e at the most. On the other hand, the historical context clearly requiresa large Hebrew population atth e dramatic transition from theCanaanit e period toth e Israelite period. The food provisioning of so large amigratin g community in the hyper-arid Sinai desert is far beyond the regional carrying potential. Something"ver y extraordinary must have occurred indeed, about which theScripture s are very definite andclear .

187 12 Relationships between the landscape, climatic, and agricultural history of the region

VALLEY CHANGESI NRELATIO NT OCLIMAT EAN DAGRICULTUR E

INTRODUCTION In the sphere ofhistori c landscape studies the assessment of possible causal relationships between valley history, climate history and land-use history, should firstan dforemos tb egrounde do nth eabsolut e basis of stratigraphy: TIME. Only timei sth eimpartia l dimension that allowsfo ra juxtapositiono fvariou s aspectso fth e historyo feart han d man. Ifther e appearst ob ea connectio n between climatic historyan d landscape history,o nth ebasi so ftime ,i tdoe sno t necessarily mean thatw eunderstan d this relationship interm so fclimati can dgeomorphi c processes. The extrapo­ lation ofknowledg e about present climatican dgeomorphi c processesma y vice versa lead to amistake n interpretationo f certain landscape properties, considered to bei nequilibriu m with thepresent , buti nfac t formed in a different environmento fth e past, orove ra lon g periodo ftime . The ageo f the landscape properties under consideration oughtt ob e determined before relations aremad ewit h present processes. The dimensiono ftim e should never be underestimated and the present doesno talway shol dth e proper keyt o unravel thepast . The influenceo fma ni sbeyon d doubt with regardt oth ewadi s terracedfo r runoff farming purposes, inwhateve r historical period. Iti sclea r thatma n had a decisive influencei nthi s respecto ngeomorphi c wadi processes. By building thousandso fcheck-dam sacros s hundredso fwadis , mandi d putth e breaks onth e occasional runoff floodwaters that turna dr yrive rbe dint o a

188 stream in the desert on some rainy days during the year,usuall y in the winter season. Aggradation took place in the terraced wadi courses as aresult , as shown in the examples ofHorva t Haluqim and NahalMitnan . There is no need here to invoke climatic causes. Man did the job that caused aggradation, although (!) incertai n historic periods more sediment may have been supplied from the catchment toth e wadis than in other periods,lik e today. The possible influence ofma n is,however ,muc h more difficult toprove ,i f at all, when acut-and-fil l stratigraphy is found in avalle y that lacks a staircase of runoff farming terraces in its stream bed. The investigated Kadesh-Barnea valley isa n excellent example in thisrespect . During the Late Bronze and Iron Age, aswel l as during the subsequent Persian and Hellenistic periods, no discernible evidence was discovered about sweeping changes in the valley ofKadesh-Barne a and its stream channel.Th ewad i bed was apparently at about the same level astoda y or somewhat lower, in a kind of long-term dynamic equilibrium, not much fill and not much cut.Th epossibilit y is always there,o fcourse ,tha t significant changes did occur during those periods,bu t either left no trace for detection today or were overlooked by the author.

THECLIMATI C RELATIONSHIP The studied cut-and-fill history of theKadesh-Barne a valley is in the following section first compared with the climatic history,juxtapose d asgoo d aspossibl e on the basis oftime :

ROMAN FILL DEPOSIT. The first detectable aggradational period during the last three millennia predates theLat e Byzantine period. Aggradation may have taken place inRoma n times,a s indicated by aC-1 4dat e (Figure 29,Tabl e25) . The precise beginning ofthi s aggradational phase could not be determined, however, with regard to regional paleo-climaticdata ,i t isnoteworth y that a dramatic rise in the level of theDea d Sea occurred during the 1st century B.C., which isunderstoo d to reflect an increase in rainfall (Klein, 1982).A time-relationship between this rainy period and pre-Byzantine aggradation in theKadesh-Barne a valley cannot be substantiated, asth edatin g ofth e Roman fill lacks sufficient detail.

INCISION DURING THEEARL Y ARABPERIOD . Downcutting tookplac e after the 7th century A.D. and prior to the next aggradational period, which apparently began around 1200A.D .Wit h regard toclimate ,Lam b (1977:428)quote s evidence of widespread drought inth e eastern Mediterranean during the 7th and 8th

189 centuries.

MAIN FILL DEPOSIT (LATECRUSADE R -MAMLU K -EARL Y OTTOMAN PERIOD). The most remarkable period ofaggradatio n inth e Kadesh-Barnea valley during historical timesoccurre d from about 1200-1700A.D. ,whe nprogressiv e sedimen­ tation caused the levelo fth ewad i bed to risewit h some4 meter . It isa grea t advantage that well-dated and detailed proxy-climatic rain­ falldat ao f localorigi nexis t over thisperio d oftime .Thes e datahav e been derived from variations inth e level of theDea d Sea (Klein, 1981) and the width of tree rings from aJuniperu sPhoenic a tree (Waisel and Liphschitz, 1968). Thistree ,datin g back from 1115-1968,gre w just4 0k m west ofKadesh - Barnea on GebelHalal , ata simila r altitudeo fabou t40 0m a sth e Kadesh- Barnea valley. Thedat a from the tree ringsan d Dead Sea level support each other very well indeed (Klein, 1981). From 1170-1700th eclimat ewa s generally wetter than usual, especially in theperiod s from 1170-1320, 1360-1430, and from 1470-1650 (Klein, 1981). It isnoteworth y thatman y charcoal samplescollecte d forcarbon-1 4dating , date back to the intervenient period between 1320-1470, when the climate was lesswet , alternating with drier periods.Perhap s therewa s apaus e inaggra ­ dation during thisstage . The valley surface near the stream channel may have remained unchanged asa resul t forsom e 150years , enabling relatively more asho rcharcoa lt odevelo pupo nit .

INCISION SINCE THELAT EOTTOMA N PERIOD UNTIL TODAY. Rapid incision lowered the wadi bed again with some4 mete r during the lasttw oo r three centuries. From 1700-1880th eclimat ewa sver y dry, but improved somewhat at the end of the 19th century and during the first decades ofth e20t h century. Several periods of drought have since occurred inth e rather dry present century (Klein, 1961,1981) .

This pattern ofaggradatio n and incision during historical times in the Kadesh-Barnea valley appearst ob emor ecomplex , asa detaile d case study, than the resultspublishe d by Vita-Finzi (1969)abou t the widespread occur­ renceo f ahistori c filli n theMediterranea n valleys ingeneral . Theprinci ­ pal difference istha ta perio d of incision during theEarl y Arabperio d seems to separate the historic fill into twoparts : 1)A Roma n fillo fmoderat eproportions . 2)Th emai nhistori c fill,deposite d from about1200-1700 .

190 The results from theKadesh-Barne a valley do,nevertheless ,corroborat e the basic trend of Vita-Finzi'spioneerin g research in this field, as far as historical aggradation isconcerned . Although various valleys may have their own specific pattern ofcut-and-fil l processes, thewidesprea d occurrence of stream valley aggradation in the Mediterranean region and the Near East around the time of theMiddl e Ages seems to suggest a general climatic cause, also favoured by Vita-Finzi (1969:115). The influence ofma n cannot be ruled out (Davidson et al., 1976;Butzer , 1980;Wagstaff , 1981)an d may be the dominant factor incertai n cases, e.g. the terraced wadis inth eNegev . Each valley should be examined within itsow n sphere and upon itsow nmerits . Concerning the valley ofEi n elQudeirat , the conclusion seems justified that the influence ofclimati c variations appears tob eth epredominan t cause of thecut-and-fil l history of thisparticula r valley. Periods of aggradation coincide with a climate that iswette r than usual,wherea s periods of incision seem to occur during particularly dry climatic periods. This conclusion veri­ fies theworkin g hypothesis adopted by Goldberg (1984),base d upon his studies of theLat e Quaternary valley stratigraphy inth eKadesh-Barne a area,i n which he concentrated on theLat e Pleistocene.

ASSESSMENT OF POSSIBLE ANTHROPOGENIC INFLUENCES ON THE CUT-AND-FILL HISTORY OF THE KADESH-BARNEA VALLEY Changes in land-use did occur inthi s remote and very thinly populated desert area,bu t the juxtaposition ofth e cut-and-fill evolution of the valley and the history of runoff farming settlement periods inth e Negev (Israelite period,Lat e Nabatean-Roman-Byzantine-Early Arab period)doe s not seem to make much sense. It seemshighl y unrealistic to explain theperiod s of aggradation and incision in the valley ofEi n el Qudeirat as theresul t ofman . On the first facea tentativ e coincidence might becontemplate d between the poorly dated Roman fill and thepossibl e onset of runoff farming in the catchment area ofNaha lKadesh-Barne a inNabatean-Roma n times. However, there is so far no evidence of settlement in the area during Nabatean-Roman times, as most historic archaeological remains in theare a belong primarily to the Bronze and Ironag e orth eByzantine-Earl y Arabperio d (Haiman, 1982,persona l communication), which is substantiated by the outcome of the C-14 data published in thisthesis . Moreover, historical documents studied by Mayerson (1963)see m to indicate that there was no runoff farming south-west ofShivt a in the late 4th century A.D. The settled area oroecumen eha d itsboundar y inthos e days near Shivta,

191 whilst the inner desert began south-west ofShivta , according to the narra- tiones ofNilu sth eAscetic . By theLat e Byzantine period the limit of the inhabited part ofth ecentra l Negev desert had extended itselfwes t and south for a considerable distance,a s indicated by the itinerarium (ca.A.D . 570)o f Antoninus ofPlacenti a and the Nizzana papyri (Mayerson, 1963). The carbon-14 results ofth e described aquaduct remains in theKadesh-Barne a valley, dating to the Bronze age and 7th century A.D., give additional support to the view that post-Iron age settlement and agricultural land-use ofthi s area south­ west ofShivt a mainly began in theLat e Byzantine period. Most authorities agree that the sedentary runoff farming civilization in the central Negev and north-eastern Sinai ceased toexis t during the Early Arab period, when towns and farms were abandoned because ofpolitical ,econo ­ mic and religious changes. How should this have led to incision in the valley ofEi n elQudeirat ?A possibl e deterioration of runoff farming terraces in the various tributaries ofNaha l Kadesh-Barnea upstream from the Valley ofEi n el Qudeirat, might have supplied more sediment, perhaps causing aggradation downstream rather than incision? Let us continue, for the sake ofargument , to consider possible human causes for aggradation or incision in later periods aswell . What could man have• don e in thecatchmen t of Nahal Kadesh-Barnea during the Late Crusader period, to cause theonse t of themos t remarkable period of aggradation in historical timesan d tomaintai n it forman y centuries?Wa s there any dramatic change in land-use during those days in the above catchment? There is no indication in that direction. In which way did man behave differently in the catchment of Nahal Kadesh-Barnea during the later part of the Ottoman (Turkish) period, that might explain the termination of this extraordinary aggradational period and the start of rapid incision? Moreover, it seemswort h noting,tha t the significant changes that followed theFirs t World War (theBritis h Mandatory period and theestablishmen t of the State of Israel)di d not cause any notable alteration in the current phase of incision and erosion, notwithstanding the building ofth e Jarvis dam in the Kadesh-Barnea valley and its functioning from 1926-1969.

REFLECTIONS ABOUT THERELATIO N BETWEEN CLIMATE AND VALLEY HISTORY How then should a relatively wet climatic period lead to aggradation? Although the time-stratigraphic relation seems irrefutable, it appears rather difficult to explain this relation in terms of landscape processes. More rainfall may lead toa differentia l increase in vegetation: a slight increase

192 on the hillside catchment and a much more dramatic increase in the stream channels, due to thedifferen t soil depths and moisture storage capabilities on the hill-slopes and valleys,respectively . Such a development might help to slow down silt laden runoff streams in the wadis, thereby perhaps inducing aggradation. This is, however, but one possible aspect of thecomple x and intricate reaction of the landscape tomor e rainfall. Another important question is to what extent the rainfall regime varied in wet and dry periods, with regard to rainfall intensity and duration? A certain amount ofrai n may come down in many light showers or in a few heavy downpours, which hasa rather different effect in terms of infiltration, runoff,erosion ,an d length of stream flow in the drainage system. Did the aggradational sediments only originate from soil stripping of the hill-slopes in the catchment, or was there amor e intensive dust input from outside the region as well?Durin g thewe t climatic periods of the Late Pleistocene large amounts of loess were deposited in theCentra l and Northern Negev (Yaalon and Dan, 1974; Bruins,1976 ;Bruin s and Yaalon,1979 , 1980; Issar and Bruins, 1983). Isther e any similarity inclimati c and landscape processes between the wet climatic periods in Historical and Late Pleistocene times? As climatic processes ought tob estudie d ina wider geographical context, the possible relation between Southern Levantine conditions during the last IceAg e in the Late Pleistocene and during the so-called Little IceAg e inhistorica l times deserves more attention. There occurred a general turn towards colder in , according toLam b (1977:449; 1984), from A.D.* 1200-1400 onwards until the 18-19th century. This was accompanied by shiftso fth e zones of most cyclonic activity, as the polar cap and the circumpolar vortex expanded. It seems clear that during a period of aggradation more sediment is sup­ plied to the landscape drainage system than itca n handle. The sediment thus accumulates in thewadis , whereas during aperio d of incision sediment is eroded from the stream channels and subsequently removed. The possibility that incision and erosion inth euppe r part ofth e drainage system may cause contemporaneous aggradation in the lower part of the drainage system, is a further complicating factor that requires attention, not just in terms of single events but also in stratigraphic terms over long periods oftime . More research isneede d to unravel the intricate processes that cause aggradation and incision in relation towe t and dry periods, respectively.

193 CLIMATICVARIATION SAN D THEVIABILIT Y OFRUNOF F FARMING

PALEO-CLIMATE INTH EREGIO N DURINGHISTORICA L TIMES Climate expresses the integrated atmospheric stateo fou r planet in time and space. Thestud y ofchange si n climate and itspossibl e impact on human society,b ywhic h therelatio n of climatet o foodproductio n isprobabl y most important,i snecessaril y tied totim ean d acertai ngeographi c locality. As instrumental recordso fclimat e are limited toth elas t centuries only, paleo-climatic dataca nonl y be obtained indirectly from proxy-climatic indi­ cators, likehistorica l documents, changes in lakelevels , and thewidt h of tree ringsdate d by dendrochronology. Information deduced from theBibl e aboutclimati ccondition s in theLan d of Israel and its surroundingsdoe sno t suggest sweeping long-term changes since the timeo fAbraha m until the 1st century A.D. Thegenera l fabric of climate has apparently remained rather similar during thisperio d oftime . From the perception ofhuma nexperience , however, drastic short-term climatic varia­ tionssee m tohav eoccurre d within Biblical times, liketh e faminedurin g the MiddleBronz e agei nth eday so fJaco b and Joseph forexample . It isa matte r of semantics and time-scalewhethe r tocal l theseclimati c events changes, variations, or fluctuations. The iceage sar eals omer e fluctuationso f the climate ofth eeart hwithi n geological or astronomicaltimes . Proxy-climatic data from tree rings prior to 1115 A.D (Waisel and Liphschitz, 1968)ar eapparentl y not available yet.Relativel y high levelso f the Dead Sea, implying awette r than averageclimate ,appea r tohav eoccurre d in historical timesdurin g part ofth eBronz eAge , although reliable dating seemst ob e aproble m (Neev and Emery, 1967;Nee v and Hall, 1977). Subsequent medium to long-term periodswit h awette r than averageclimat e in historical times occurred in the 1st century B.C., as indicated by a sharpris e and fallo f theDea d Sea level (Klein, 1982), and particularly during the rather unique climatic period from about 1170-1700A.D . (waiselan d Liphschitz,1968 ; Klein, 1981).

RELATIONSBETWEE NCLIMAT EAN DRUNOF F FARMING Considering the possible relationsbetwee n thesewette r than average climatic conditions and thehistor y ofrunof f farming,ther e isa bi t ofiron y in the facttha t thisuniqu e form of arid zone agriculturewa sno t practised

194 in the Negev by a sedentary population when theclimati c environment seemed to have been most ideal, i.e. during the 1st century B.C. and from A.D. 1200- 1700. In other words, there isn o apparent relationship between long-term climatic trends and runoff farming settlement patterns inth e central Negev desert during the last three millennia. The viability of runoff farming does not only relate to long-term climatic trends, but is also very much dependent upon annual rainfall variability and the frequency ofdrough t years. Runoff farming enables agricultural production in certain arid regions based upon local rainfall.Yet ,runof f farmers have to cope with drought years, as arid regions are characterized by large yearly fluctuations in the amount of runoff producing precipitation. How did the ancient farmers and the sedentary population inth ecentra l Negev in general cope with annual rainfall variations and droughts? In the present stateo fknowledg e we can only guess about annual data on rainfall variability and droughts during the twodistinc t historic phases of sedentary runoff farming in the central Negev, i.e. the Iron Age and the composite Late Nabatean-Roman-Byzantine-LateAra b period (Mayerson, 1955; Evenari et al., 1958;Aharon i et al., 1960;Cohen , 1976;Negev , 1982a, 1982b; Haiman, 1982). The data acquired since 1960a t the three experimental runoff farms of Avdat, Shivta, and WadiMashash , established by Evenari, Shanan and Tadmor (Evenari et al., 1971, 1982)giv e valuable information about annual rainfall and runoff variations inpresen t times. On the basis ofthes e available data a rough estimation wasmad e about the frequency ofdrough t years, in which food production would probably have been a failure (Table 27, Bruins et al., 1986a,b). More or lesssimila r frequencies of drought years probably formed a part of normal climatic patterns in the past during similar dry-to-average climatic phases,a s in theperio d from 1960-1984. The pioneering research about runoff farming by Evenari and his colleagues was not designed to investigate long-term viability with respect to food crops in a self-sufficient food economy. Their data can,therefore ,onl y serve asa basis for additional and different investigations, specifically designed to study the yields of food crops in relation to rainfall,runoff ,soi l fertility and various cropping patterns on a continuous long-term basis. Such research can give more detailed and firm answers about theviabilit y of food production through runoff farming ina dry arid climate on the border of the hyper-arid zone. With regard to the rainfall and runoff data from the farms mentioned above,

195 it iseas y to distinguish between a very rainy year and aba d drought year.I t is difficult, however, todra w the dividing linebetwee n amoderatel y good year, in which agricultural production isperhap s just above the limit of economic or nutritional survival and amoderatel y bad year, just below that red line. Another aspect of viability, that cannot be grasped in terms of overall annual rainfall and runoff data, isth e distribution of the incoming runoff water over the farm fields. At theAvda t farm, forexample , largely based on thecondui t system, a choice can bemad e out ofsevera l options as to how many fields should receive the incoming runoffwater . Inver y rainy yearsal l the fieldsca n be saturated with runoff water. But in moderate toba d years some fieldshav e to beclose d off for the incoming runoff waters toensur e that at least part of the cultivable area is saturated with moisture,i fa t all.A n even water spreading over all the fields,i n such a situation, could result in a total crop failure. Theterrace d wadi system, as found at HorvatHaluqi m for example, seems tob e less flexible in this respect. Thehydrologica l and agricultural functioning ofthi s latter system has not been investigated so far and different catchment properties preclude simple and possibly misleading comparisons to bemad e on the sole basis of runoff water management flexibility. The timing and frequency ofrunof f floods during the growing season is another significant climatic factor ofdirec t importance in viability studies. The season 1964-65wa s a very rainy year at Avdat (144.2 mm). Since the first flood came late, however,i t was decided by Evenari and hiscolleague s (1968) not to sow wheat, butothe r field crops and vegetables, like chick pea, carrots, asparagus and artichokes. There aren o dataho w the ancient runoff farmers selected their field crops with regard toth e timing of the first flood.Neithe r dow ehav e such detailed paleo-climatic data about rainfall and runoff distribution in thecentra l Negev in antiquity.

THEFREQUENC Y OF DROUGHT YEARS The modern rainfall and runoff data ofth e three experimental runoff farms enable an assessment of the frequency of runoff drought years at present.A runoff drought year usually coincides with a rainfall drought year, though not necessarily. More runoff is sometimes produced inyear s with less rainfall than in wetter years, because of annual differences in rainfall intensity and duration. Detailed runoff data for theAvda t and Shivta farms are only available from 1960-61 until 1966-67. Although the runoff data from the Mashash farm are less precise, as they were measured in a simple way,

196 TABLE27 . Runoffdrough tyear sa tth eAvdat ,Shivta ,an dWad iMashas hexperi ­ mentalrunof ffarms , basedupo navailabl erainfal lan drunof fdat a (afterBruin se tal. ,1986a,b) .

AVDAT RUNOFF FARM SHIUTA RUNOFF FARM MASHASHRUNOF F FARM hillsid econdui t hillsidecondui t liman system system system

RAINY Rain Runoff Drought Rain Runoff Drought Rain RunoffDrough t SEASON (mm) (m3) year (mm) (m3) year (mm) (m3/ha) year

1960-61 70.0 1932 102.5 594 1961-62 64.7 4471 50.6 280 * 1962-63 29.5 792 * 24.8 44 • 1963-64 183.3 51875 145.3 6293 1964-65 144.2 6115 160.8 9444 1965-66 91.5 6905 67.1 1297 1966-67 80.7 1827 94.4 2285 1967-68 86.9 75.4 1968-69 70.3 73.2 1969-70 58.4 * 43.6 * 1970-71 82.0 115.2 142.0 7000 1971-72 162.6 172.4 207.0 10000 1972-73 53.5 * 41.0 * 55.0 0 » 1973-74 132.6 105.8 180.5 4000 1974-75 84.9 77.7 93.0 3000 1975-76 77.1 62.6 * 79.5 0 * 1976-77 64.1 * 101.0 2000 1977-78 68.0 75.0 1500 1978-79 56.0 * 90.7 5000 1979-80 100.6 166.0 16500 1980-81 100.8 120.7 10000 1981-82 68.8 # 75.5 0 * 1982-83 124.2 170.0 3500 1983-84 29.3 * 84.5 0 *

Average 86.8 10560 88.3 2891 117.2 4464

Droughtyea r frequency 3.4 3.2 3.5 (onceever y ...years )

their consistentyearl ydeterminatio ni so fgrea tvalu et o assess long-term viability. Data from the Avdatrunof ffar msugges ttha tapproximatel y 7ou t of 24 yearsca nb eclassifie da sdrough tyears .Thu sth eaverag efrequenc yo frunof f droughts at theAvda tfar mi sonc ei n3. 4years . The Shivta runoff farm, another example of acondui tsyste mwit hpossibilitie s to manipulate the

197 incoming runoff waters, experienced about 5drough t years during 16year s of measurements, an average drought frequency ofonc e in3. 2 years. At theWad i Mashash runoff farm, largely based on the liman system fed by larger catch­ ments which show clear zero runoff yields indrough t years (Table 27), 4 out of 14 years were absolute runoff drought years, an average of once in 3.5 years. As there appears a remarkable similarity between the resulting figures from the three runoff farms,th epreliminar y conclusion seems justified that in the 80-120 mm winter rainfall zone of the central Negev, shortage of runoff water seems a severe limiting factor in agricultural production once every 3 to 4 years in the present rainfall regime.Fro m a farm management point of view,i t isobviousl y a must to overcome these runoff drought years ifrunof f farms are to be viable. Thequestio n immediately arises how individual farmers and society as awhol e managed tocop e with runoff drought years in antiquity.

DROUGHTS AND THEVIABILIT Y OF RUNOFF FARMING It should first ofal lb e realized that, apart from zero yields, it is difficult to define adrough t in absolute terms. A drought in agricultural production isassociate d with acertai n economic and socio-technical framework within a certain society. Apart from natural environmental factors,th e fabric of that society and theamoun t ofpeopl e that depend upon a certain area of agricultural land for their provisioning, determines when a drought threshold iscrossed , so that the system can no longer function in the usual way for a certain time. When thrown out ofequilibriu m because ofdrought ,th e system tends to draw on internal reserves of energy, and/or resorts to external energy that actually forms the reserve ofa socio-economic orpolitica l framework wider than the system initially described. The restored balances and their thres­ holds of tolerance areno t necessarily the same asth e initial balances or thresholds of tolerance (Spitz, 1980). The forming of internal reserves was probably a significant strategy of the sedentary population that engaged in runoff farming toovercom e drought years during the Israelite period, as well as during the composite Late Nabatean- Roman-Byzantine-Early Arab period. Reserves could havebee n accumulated in three principal commodoties: water, food,o rmone y (Bruins et al, 1986b). It seems unlikely that any significant quantities ofwate r might have been stored in wet years for irrigation in dry years. The capacity of the available cisterns seems to have been sufficient only for direct water requirements of

198 man and his animals,no t for agriculture,wit h possible exceptions like one of the cisterns atHorva tHaluqim . Even in the latter case, water was probably not stored for agricultural use indrough t years, but for supplementary irri­ gation of a few fields in the same year only. During drought years, all water reserves wereprobabl y necessary to sustain man and hisanimals . It remains a matter ofuncertaint y whether these stockpiles of food or money could have been acquired through runoff agriculture in the central Negev desert during thegoo d years as a reserve for the drought years, thereby maintaining a self-sufficient equilibrium. There are several archaeologic (Cohen, 1976, 1980) and historic indications (Gihon, 1967)tha t a central government outside theNegev , to which the latter region belonged, had an interest inestablishin g andmaintainin g a sedentary population inth e central Negev desert for defensive purposes, both during the Israelite and Roman- Byzantine period. In that perspective it seems likely that thecentra l government and its economic power constituted the principal buffer onwhic h the sedentary popula­ tion of the central Negev could rely for itsprovisionin g intime s ofdrought . The imperial government, ofwhic h the central Negev wasa par t during Roman and Byzantine times, apparently had a clear interest in the activity of runoff farming in; the area and probably gave guarantees for food supplies and/or financial compensation in years ofdrough t when crops failed. Runoff farming waspractise d forman y centuriesdurin g thecomposit e Late Nabatean-Roman-Byzantine-Early Arab period, and the system obviously func­ tioned despite the inevitable occurrence of droughts. The integration of the region in a larger political and economic order must have been the key to overcome the drought years.A n assured supply of food (and water if necessary) must have been guaranteed to some extent as an incentive to settle the region. Such abackgroun d doesno t rule out, however, the development of complex economic relations within the runoff farming district itself. The basic prin­ ciple of economic integration in a larger political setting during Roman- Byzantine times asth eke y of viability seems indisputable, because in no way could the runoff farming district haveproduce d enough food to sustain its population, as indicated inth e following and final parto f this dissertation. However, in transitional periods and other special cases, when the wider political and economic integration collapsed, individual farmers might have been able to operate ina self-provisioning way. This important matter of self-sufficient runoff farming viability in such a dry area as the central Negev desert requires further investigation.

199 FOOD PRODUCTION ANDPOPULATIO N INTH ERUNOF F FARMING DISTRICT

In historical timesth eNege v wassettle d bya sedentar y population that engaged in runoff farming during the Israelite period and during the composite Late Nabatean-Roman-Byzantine-Early Arab period (Mayerson, 1955; Evenari et al., 1958;Aharon i et al., 1960;Cohen ,1976 ;Negev , 1982a,1982b ; Haiman, 1982). Thepossibl e existence of runoff farming communities during the Bronze Agecanno t berule d out, but the picture is far from clear (Cohen and Dever, 1980). The survey carried out byKeda r (1967)determine d the extent of the runoff farming district inth ecentra l Negev (about200 0squar e km), as well as thetota l areao fterrace d fields (about 4000hectare) . It is not known, however,ho wmuc h wadi land was actually cultivated within each histo­ rical period. There seems tob ea genera l consensus amongst scholars that the sedentary population of thecentra lNegev , aswel l asth e extent of runoff farming, reached itspea k during the Byzantine period. Thepopulatio n level in this period wasestimate d byAvi-Yona h (1964; qf. Broshi, 1980)t o have ranged between 52,000-71,000 people. Broshi (1980)arrive d ata mor e conservative figure. In his opinion thepopulatio n ofth ecentra l Negev did not exceed 30,000peopl e during Roman-Byzantine times. Thequestio n arisest owha tdegre e thesedentar y population ofth e central Negev desert could havebee n self-sufficient interm so f food production. In order toapproac h thismatte r ina quantitativ e manner, it is first necessary tomak e anumbe r ofsupposition s that areno t entirely realistic but, never­ theless,require d asa basi s forcalculations :

1)Al lth e400 0h ao fterrace d wadi fields were used forrunof f farming during Byzantine times. 2)Al l farming land wasuse d for wheat production. 3)Th e average yield ofwhea t amounted to some60 0kg/hectare , as extensively discussed inchapte r 7. 4)Th eaverag e annualbul k food requirement percapit a isequa l to about 250 kg ofwhea t (Broshi,1980) , discussed inchapte r11 .

Based on thesesuppositions ,i t can be calculated that annual wheat produc­ tion in the runoff farming district wasprobabl y not higher than about 2,400,000 kg. Such anamoun t ofwhea t would havebee n sufficient to feed some 9600 people. Ifsomeon e would consider an average annual yield of60 0 kg/ha

200 too low, let's be optimistic and base our calculation on a runoff-farming bumper harvest, for those days, of 1000kg/ha .Th eresultin g 4,000,000 kg of wheat could have fed some 16,000 people, which isstil l only half the conser­ vative population estimate made by Broshi (1980). Moreover,thi s latter wheat- yield figure seems highly unrealistic asa n annual average, because the drought years must also be taken into account. Even the lower figure of2,400,00 0k g wheat hast o be reduced by the quantity of seed needed for next year (about 400,000kg )an d some 10 %(Hall , 1970)o fpost-harves t losses (about 240,000 kg), which leaves some 1,760,000 kg for human consumption. This quantity would be sufficient to feed about 7000 people. The inclusion of the Kadesh-Barnea valley and therunoff-farmin g areas in north-eastern Sinai, which belonged to thedistric t inByzantin e times, might have added food for another 1000peopl e (Chapter 11). As it is unlikely that all available farm land could have possibly been devoted to wheat farming each year,i t can safely beassume d that less than 8000peopl e could have been fed in the runoff farming district in a self-sufficient food economy. This means that less than a quarter of the conservative population estimate of 30,000,mad e by Broshi (1980), could have been fed by locally produced food in the runoff farming district. The conclusion seems therefore justified, that runoff farming during theByzantin e period could nothav e matched the direct food requirements of the sedentary population inth ecentra l Negev. An assured food supply to thispopulatio n from outside the region must have been an integral part ofpolitica l and economic realities inByzantin e times. Local runoff farming seemed tohav e played a supplementary function in this respect. However, during changing, transitional phaseso f sedentary agriculture in the runoff farming district, e.g. inEarl y Arab times, it is nevertheless possible that the remaining farmers,strugglin g tocop e with the new political and economic situation, were able to subsist on their ownproduce . Also in other periods with low population levels, a kind of self-sufficiency might have been attained. This latter aspect requires further research.

201 APPENDIX

ARCHAEOLOGICALAN DHISTORICA LPERIOD SI NISRAEL ,WIT HEMPHASI SO NTH ENEGE V

Period Approximatedat eB.C .o rA.D .

NEOLITHICUM 7500- A00 0 B.C.

CHALCOLITHICUM 4000- 320 0 B.C.

CANAANITEo rBRONZ EAG E EarlyBronz eAg e 3200- 220 0 B.C. MiddleBronz eAg e 2200- 155 0 B.C. LateBronz eAg e 1550- 120 0 B.C. ISRAELITEo rIRO NAG E EarlyIro nAg e 1200- 102 0 B.C. MiddleIro nAg e 1020- 842 B.C. LateIro nAg e 842- 587 B.C.

BABYLONIANAN DPERSIA NPERIOD S 587- 332 B.C.

HELLENISTICPERIO D 332- 37 B.C.

Inth eNegev :

NABATEANPERIO D* EarlyNabatea nPerio d 300- 100 B.C. MiddleNabatea nPerio d 25BC -5 0 A.D. LateNabatea nPerio d 70- 106+A.D . Annexationo fth eNabatea nkingdo mt oth eRoma n empire 106 A.D.

ROMANPERIO D 37BC-32 4 A.D.

BYZANTINEPERIO D 324- 634 ,,

ARABPERIO D 634- 109 9

CRUSADERPERIO D 1099- 129 1

MAMLUKPERIO D 1291- 151 6

OTTOMANPERIO D 1517- 191 7

BRITISHMANDATOR YPERIO D 1917- 194 8

STATEO FISRAE L 1948-

*Base do nNege v (1982a).

203 APPENDIX

LABORATORYMETHOD SO F SEDIMENT AND SOIL ANALYSIS

Particle sizeanalysi so f thesedimen t and soilsample swa scarrie d out at the Stichting Technisch Centrum voor deKeramisch e Industrie in Arnhem. The samples were treated withHC1 , so that thecarbonat e components were removed prior toth e grain sizedeterminations . Some largecarbonat e grainsma y not havebee n dissolved entirely. All other chemical analyseswer ecarrie d outb yMr . F.J. Lettink at the Chemical Laboratory of the Department of Soil Science and Geology (Agricultural University ofWageningen) , headed byMr . L. Th.Begheijn . The methods for the quantitative determination of organic carbon, nitrogen, carbonate, soluble salts, electrical conductivity and pH are described in Methods ofChemica lAnalyse s forSoil san d Waters (Begheijn,1980) . Themetho d used to determine thetota lphosphat e content isdescribe d by Begheijn and Schuylenborgh (1971:136). The pH, soluble saltsan d electrical conductivity weremeasure d in the liquid extract ofa 1:1 soil/water solution. Allanalyse swer e carried out on the natural soil or sediment fraction <2 mm . Incas egrave l (>2 mm) was found tob epresent , theweigh tpercentag e ofth egrave l fractionwa s deter­ mined,whic h includesth ecarbonat e grainsa swell . Large size thin sections forth emicroscopi c investigation of soil, sedi­ ment and ancient aquaduct-mortar samples weremad ea t the Micromorphological Laboratory of the Department ofSoi l Science and Geology (Agricultural University ofWageningen) , headed by Ir. R. Miedema.Th ethi n sectionswer e prepared byMr .O.D . Jeronimus,accordin g toth emetho d described by Jongerius and Heintzberger (1963).

204 REFERENCES

ABEL,M.E. ,SCHNITTKER ,J.A .an d BROWN,D.C . (1981)Climate-defensiv e policies toassur e food supplies. In:Climate' s Impact onFoo d Supplies (L.E.Slate r and S.K. Levin, eds). AAAS Selected Symposium 62, pp. 45-60. Boulder, Colorado:Westvie wPress . AHARONI, Y., EVENARI, M., SHANAN, L. and TADMOR,N.H . (1960)Th e ancient desert agriculture of the Negev. V. An Israelite settlement at Ramat Matred. IsraelExploratio n Journal 10:23-36,97-111 . ARKIN,Y .an d BRAUN,M . (1965)Typ e sections of upper Cretaceous formations in the northern Negev (southern Israel). Geological Survey of Israel, Stratigraphic sections,N o2a . ARZI, A. (1981)Soil-Landscap e relationships inth e hilly area of SdeBoqer . M.Sc. Thesis,Bar-Ila n University (Israel),Dep .o fGeograph y (in Hebrew). ASHBEL, D. (1973)Climate . In: Geography, pp. 94-111. Jerusalem: Keter Publishing House. AVI-YONAH,M . (1964)Essay s and Studies.Jerusalem : Newman (in Hebew). BAILEY, C. (1980)Th eNege v in the nineteenth century: reconstructing history from Bedouin oral traditions. Haifa: The Institute of Middle Eastern Studies: Asian and African Studies14:35-80 . BARTH, F. (1959/60)Th e land use pattern ofmigrator y tribes of South Persia. Norsk Geografisk Tidsskrift 17 (The Bobbs-Merrill Reprint Series in the Social Sciences,A-ll):l-ll . BARTH, F. (1964) Nomads of South Persia. TheBasser i Tribe of the Khamseh Confederacy. Oslo:Universitetsforlaget . BART0V, Y. (1974)A structura l and paleogeographic study of the Central Sinai faultsan d domes.Ph.D .Thesis ,Th e Hebrew University of Jerusalem,Depart ­ ment ofGeolog y (inHebre w with English abstract). BART0V, Y., EYAL, Y.,GARFUNKEL ,Z . and STEINITZ,G . (1972)Lat e Cretaceous and Tertiary stratigraphy and paleogeography of southern Israel. Israel Journal ofEart h Sciences21:69-97 . BART0V, Y. and STEINITZ, G. (1974)Th e Judea and Mount Scopus Groups in the Negev and Sinai with trend surface analysis of the thickness data. Israel Journal ofEart h Sciences26:119-148 . BEGHEIJN, L.Th. (1980) Methods of Chemical Analysis for Soils and Water. Wageningen:Agricultura l University,Departmen t of Soil Science and Geology (third edition). BEGHEIJN, L.Th. and SCHUYLENBORGH, J. VAN (1971)Method s for the Analysis of Soils. Wageningen:Agricultura l University,Departmen t of Soil Science and Geology. BENJAMINI, C. (1979a)Facie s relationships of the Avdat Group (Eocene)i n the northern Negev. Israel Journal ofEart h Sciences28:47-69 . BENJAMINI, C. (1979b)Th eNaha l Yeter Formation -a ne w Eocene formation from the northwestern Negev, Israel. Israel J. of Earth Sciences28:164-166 . BENJAMINI, C. (1980)Stratigraph y and foraminifera of theQezio t and Har Aqrav Formations (latest Middle toLat e Eocene)o f thewester n Negev, Israel. Israel Journal ofEart h Sciences29:227-244 . BENJAMINI, C. and ZILBERMAN,E . (1979)Lat e Eocene coral reefs of the western Negev, Israel. Israel Journal of Earth Sciences28:42-46 . BENT0R, Y.K. and VR0MAN, A. (1963)Th e geological map of Israel.Ser.A , The Negev. Sheet 17:Nizzana , 1:100,000. Geological Survey of Israel. BENT0R, Y.K., VR0MAN,A. , ZAK, I. (1965)Th e geological map of Israel,scal e 1:250,000, southernsheet . BERKOFSKY, L. (1984)Rainfal l patterns in the desert.Worl d Climate Programme WCP-75. Geneva:Worl d Meteorological Organization. BOERS, Th.M. and BEN-ASHER, J. (1982)A revie w of rainwater harvesting. Agricultural Water Management 5:145-158.

205 BOOHER, L.J. (1974)Surfac e Irrigation. Rome: FAO Agricultural Development Paper No.95 . BOWDEN, M.J., KATES,R.W. ,KAY ,P.A. ,RIEBSAME , W.E.,WARRICK , R.A., JOHNSON, D.L., GOULD, H.A. and WEINER, D. (1981) The effect of climate fluctuations on human populations: two hypotheses. In: Climate and History (T.M.L. Wigley, H.J. Ingram and G. Farmer, eds),pp .479-513 .Cambridg e University Press. BRAUN,M . (1967)Typ e sections of Avdat Group,Eocen e Formations in the Negev, southern Israel. Geological Survey of Israel,Stratigraphi c Sections 4. BREEMEN, N. VAN (1972)Soi l forming processes in acid sulphate soils. In:Aci d Sulphate Soils (H. Dost,ed) ,Vol . 1,pp . 66-131. ILRI Publication No.18 , Wageningen. BREEMEN, N. VAN (1986)Longter m chemical, mineralogical and morphological effects ofFe-redo x processes in periodically flooded soils. In: Iron in Soils and Clay Minerals (J.W. Stucki, B.A. Goodman and U. Schwertmann, eds). Proceedings NATO Advanced Study Institute.Dordrecht : Reidel. BREEMEN, N. VAN (1986)Persona l communication. Agricultural University of Wageningen, Department of Soil Science and Geology. BREWER, R. (1964)Fabri c and analysis of soils. New York: John Wiley. BRINKMAN, R. (1979) Ferrolysis - A soil-forming process in hydromorphic conditions.Wageningen : Pudoc,Agricultura l Research Report 887. BRINKMAN, R. (1986) Personal communication. Agricultural University of Wageningen, Department of Soil Science and Geology. BROSHI. M. (1980) The population of Western in the Roman-Byl'natine period. Bulletin of the American Schools of Oriental Research (BASOR) 236:1-10. BRUINS, H.J. (1976) The origin, nature and stratigraphy of paleosols in the loessial deposits of the NW-Negev. M.Sc. Thesis,Th e Hebrew University of Jerusalem (Department of Geology). BRUINS, H.J. and YAALON, D.H. (1979)Stratigraph y of the section in the desert loess of the Negev (Israel). Acta Geologica Academiae Scientia- rum Hungaricae 22(1-4): 161-169. BRUINS, H.J. and YAALON, D.H. (1980) Paleoclimatic significance of clay content and paleosols in the desert loess of the Negev. Bat-Sheva Seminar on Approaches and Methods inPaleoclimati c Research with emphasis on Aridic areas, Jerusalem and Rehovot. Abstracts,pp . 25-26. BRUINS, H.J., EVENARI, M. and NESSLER, U. (1986a) Rainwater-harvesting agriculture for food production in arid zones: thechalleng e of the African famine. Applied Geography 6:13-32. BRUINS, H.J., EVENARI,M . and ROGEL, A. (1986b)Runof f farming management and climate. In: Progress inDeser t Research (L.Berkofsk y and M.G.Wurtele , eds). Totowa,N.J. : Rowman and Allanheld. BRUINS, H.J, YAALON, D.H., ISSAR, A. and KARNIELI,A . (1987) The Netivot Formation, (in preparation). BRYAN, K. (1929)Flood-wate r farming. Geographical Review 19:444-456. BURINGH, P. (1977)Mondiaa l bodemgebruik en voedselproduktie. Bulletin of the Faculty of Natural Resources,Universit y of Suriname,No. l (Paramaribo). BUTZER, K.W. (1976)Earl y hydraulic civilization in Egypt:a study in cultural . Chicago:Chicag o University Press. BUTZER, K.W. (1980) Holocene alluvial seguences: problems of dating and correlation. In: Timescales in Geomorphology (R.A. Cullingford, D.A. Davidson and J. Lewin,eds) ,pp . 131-142. New York: John Wiley. CICERO. The Verrine Orations, transl.H.G . Greenwood. Loeb Classical Library. London:Willia m Heineman. COHEN, R. (1976)Excavation s at Horvat Haluqim. Atiqot 11:34-50. COHEN, R. (1980) The Iron Age fortresses in the Central Negev. Bulletin of the American Schools of Oriental Research (BASOR)236:61-79 .

206 COHEN, R. (1981)Excavation s at Kadesh-Barnea 1976-1978.Biblica l Archeologist spring-1981:93-107. COHEN,R . (1981)Archaeologica l Survey of Israel.Ma p ofSed e Boqer-East (168) 13-03. Jerusalem: The Archaeological Survey of Israel. COHEN, R. (1982)Persona l communication. Department ofAntiquities , Israel Museum, Jerusalem. COHEN, R. (1983) Kadesh-barnea. AFortres s from the Timeo f the Judaean King­ dom. Jerusalem: The Israel Museum. COHEN, R. and DEVER, W.G. (1980)Preliminar y report of the second season of the "Central Negev Highlands Project". Bulletin of theAmerica n Schools of Oriental Research (BASOR)236:41-60 . COLUMELLA. De Re Rustica,transl .H.B . Ash,E.S .Forste r and E.Heftner .Loe b Classical Library.London :Willia m Heineman. CUNNINGHAM, G.M. (1970)Waterpondin g on scalds.Journa l of the Soil Conserva­ tion Service of New South Wales26:146-168 . CUNNINGHAM, G.M. (1975)Waterspreadin g -Cobart-Byroc k district. Journal of the Soil Conservation of New South Wales31:82-94 . CUNNISON, I. (1966)Baggar a Arabs:Powe r and theLineag e in aSudanes e Nomad Tribe.Oxford :Clarendo n Press. DAM, D. VAN and PONS, L.J. (1972)Micromorphologica l observations on pyrite and itspedologica l reaction poducts. In: Acid Sulphate Soils (H. Dost, ed), ILRIPublicatio n No. 18,Wageningen . DAN, J. (1981) Soilso f the sandy region of thewester n Negev. In: Aridic Soils of Israel (J. Dan, R. Gerson, H. Koyumdjisky and D.H. Yaalon, eds), pp. 211-222.Be t Dagan: The Volcani Center,Specia l Publ. No.190 . DAN, J. (1981) Thenorthwester n part of the Negev hills and mountains. In: Aridic Soils of Israel (J. Dan,R . Gerson,H . Koyumdjisky and D.H. Yaalon, eds), pp.223-238 .Be t Dagan: The Volcani Center,Specia l Publ. No.190 . DAN, J., M0SHE, R. and ALPEROVITCH,N . (1973)Th e soils of Sede Zin. Israel Journal ofEart h Sciences22:211-227 . DAN, J., YAALON,D.H. , KOYUMDJISKY,H . and RAZ,Z . (1975)Th e soil association map of Israel 1:500,000. Bet Dagan: The Volcani Center. DAN, J. and YAALON,D.H . (1982)Automorphi c saline soils in Israel. In:Aridi c soils and geomophic processes (D.H. Yaalon, ed), pp. 103-115. Catena Supplement 1.Braunschweig : Catena Verlag. DAN, J., YAALON,D.H. , M0SHE,R . and NISSIM, S. (1982)Evolutio n of reg soils in southern Israel and Sinai.Geoderm a28:173-202 . DANIN, A. (1983) Desert vegetation of Israel and Sinai. Jerusalem: Cana Publishing House. DANIN, A., 0RSHAN, G. and Z0HARY,M . (1975)Th evegetatio n of the Northern Negev and the Judean Desert of Israel. Israel Journal of Botany 24:118-172. DAVIDSON, D.A., RENFREW, C. and TASKER, C. (1976)Erosio n and prehistory in Melos: apreliminar y note.Journa l of Archaeological Science 3:219-227. D0STAL, W. (1959) The evolution of Bedouin life. In: L'antica beduina (F. Gabrieli,ed) ,pp . 11-34.Rome . D0THAN, M. (1965)Th e fortress at Kadesh-Barnea. Israel Exploration Journal 15:134-151. DUBIEF, J. (1953) Essai sur l'hydrologie superficielle au Sahara. Alger: Gouvernement Gen. DUVDEVANI, S. (1953) Dew gradients in relation to climate, soil and topography. Desert Research Proceedings,Researc h Council of Israel Special Publication No.2 (Jerusalem). ELGABALY,M.M . (1954)Th esoils ,wate r supply,an d agriculture in northeastern Sinai. Proceedings of the Symposium on Scientific Problems ofLan d use in Arid Regions,pp . 125-152. ,Egypt . EMMETT,W.W . (1970)Th ehydraulic s of overland flow on hillslopes.U.S .Geolo ­ gical Survey, Professional Paper 662-A, U.S.A. Government Printing Office,

207 D.C. EVENARI, M., AHARONI, Y., SHANAN, L. and TADMOR,N.H . (1958)Th e ancient desert agriculture of the Negev. III.Earl y beginnings. Israel Exploration Journal 8:231-268. EVENARI, M., SHANAN, L. and TADMOR, N.H. (1963)Runoff-farmin g inth e Negev desert of Israel.Progres s report on theAvda t and ShivtaFar m Projects for the years 1958-1962. The National and University Institute of Agriculture, Rehovot, Israel. EVENARI, M., SHANAN, L. and TADMOR, N.H. (1964)Runoff-farmin g in the Negev desert of Israel (II). Progress report on the Avdat and Shivta Farm Projects for 1962/63 season. The National and University Institute of Agriculture,Rehovot ,Israel . EVENARI, M., SHANAN, L. and TADMOR,N.H . (1965)Runoff-farmin g in the Negev desert of Israel (III). Third Progress report. Avdat and Shivta Farm Projects, 1963/64season . The National and University Institute ofAgricul ­ ture, Rehovot,Israel . EVENARI, M., SHANAN, L. and TADMOR, N.H. (1968)Runoff-farmin g in the Negev desert of Israel (IV). Fourth Progress report.Avda t and Shivta Farm Projects. 1964/65, 1965/66, 1966/67 seasons. The Hebrew University, Jerusalem,Departmen t of Botany. EVENARI, M., SHANAN, L. and TADMOR, N.H. (1971, 1982)Th eNege v - The challenge of adesert .Cambridge ,Mass. : Harvard University Press. EVENARI, M., SHANAN,L . and TADMOR, N.H. (1975)Th e agricultural potential of loessan d loess-like soils in arid and semi-arid zones. Proceedings Inter­ national Seminar Land Evaluation in Arid and Semi-arid zones of . IILA,Rome ,pp .489-507 . FELIKS, J. (1963) Agriculture in Palestine in the Period of the Mishna and Talmud. Jerusalem:Magne s Press (in Hebrew). FLEXER, A. (1968) Stratigraphy and facies development of the Mount Scopus Group (Senonian-Paleocene) in Israel and adjacent countries. Israel Journal of Earth Sciences17:85-114 . FLOHN, H. and NICHOLSON,S . (1980)Climati c fluctuations in the arid belt of the 'OldWorld ' since the last glacial maximum: possible causes and future implications. In:Sahar a and surrounding seas (M.Sarntheim ,E . Seibold, P. Rognon,eds) ,pp .3-21 . Rotterdam: Balkema. FREUND, R. (1965)A mode l of the structural development of Israel and adjacent areas since Upper Cretaceous times.Geologica l Magazine 102:189-205. FROMKIN, R. (1981)Kadesh-Barnea . In: Kadesh-Barnea and its surroundings, pp..22-27. The Nature Protection Society and The SedeBoqe r Field School, Israel (in Hebrew). GANOR, E. (1975)Atmospheri c dust in Israel.Sedimentologica l and meteorologi­ cal analyis ofdus t deposition. Ph.D. Thesis, TheHebre w University of Jerusalem,Departmen t of Geology (in Hebew). GARFUNKEL, Z. (1978) The Negev -Regiona l synthesis ofsedimentar y basins. Tenth International Congress on Sedimentology (IAS), Jerusalem: Sedimento- logy in Israel,Cypru s and Turkey,Guideboo k Part I,pp . 35-110 GARFUNKEL, Z. and HOROWITZ, A. (1966) TheUppe r Tertiary and Quaternary morphology of the Negev. Israel Journal of Earth Sciences15:]01-117 . GEDDES, H.J. (1963) Water harvesting. Paper presented at the National Symposium on Water Resources,Australia n Academy ofSciences ,Canberra . GIH0N, M. (1967)Th eNege v Frontier. In: Israel and her vicinity in the Roman and Byzantine Periods (M. Gihon and S. Applebaum). The 7th International Congress of RomanFrontie r Studies. Tel-Aviv University Printing Press. GOLDBERG, P. (1984)Lat e Quaternary history of Qadesh Barnea, Northeastern Sinai. Zeitschift fur Geomorphologie N.F.28(2):193-217 . GRIGG, D.B. (1974) Theagricultura l systems of theworl d -A n evolutionary approach. Cambridge:Cambridg e University Press.

208 HACK, J.T. (1942) Thechangin g physical environment of the Hopi Indians of .Harvar d University Peabody Museum Papers 35,No .1 . HAIMAN,M . (1982)Naha lMitnan . Israel Exploration Journal 32:265-266. HAIMAN,M . (1982)Persona l communication. TheArchaeologica l Survey of Israel, Israel Museum, Jerusalem. HALL, D.W. (1970) Handling and storage of food grains in tropical and subtropical areas.Rome :FA O Agricultural Development Paper No.90 . HILLEL, D. (1959)Studie s ofLoessia l Crusts. Israel,Rehovot : Agricultural Research Institute,Bulleti n63 . HILLEL, D. (1964) Infiltration and rainfall-runoffa saffecte d by surface crusts. Bucharest: Transactions 8th International Congress of Soil Science (ISSS) 1:34-39. HILLEL, D. (1982)Nege v -Land ,water ,an d life in adeser t environment.Ne w York:Praeger . HILLEL, D. and TADMOR, N. (1962)Wate r regime and vegetation in the Negev highlands of Israel.Ecolog y43:33-41 . HOLLICK, M. (1982)Wate r harvesting in arid lands.Scientifi c Reviews on Arid Zone Research,Volum e 1:173-247. Scientific Publ., Jodhpur,India . HOPE,M . (1978)Plan t Remains. In:Earl y Arad. TheChalcolithi c Settlement and Early Bronze City. I. First-Fifth seasons of excavations, 1962-1966 (Amiran, R.), Part 3,pp .64-82 . Jerusalem: The Israel Exploration Society. HORST, L. (1983)Irrigatio n Systems. Agricultural University ofWageningen , Department of Irrigation and Civil Engineering. HORST, L. (1983)Irrigatio n systems -Alternativ e design concepts. Overseas Development Institute (ODI), Irrigation Management Network Paper 7C. HUIBERS, F.P. (1985) Rainfed Agriculture in aSemi-ari d Tropical Climate: aspectso f land-an dwatermanagemen t for red soils in India. Dissertation, Agricultural University ofWageningen . ILACO (1981)Agricultura l Compendium for rural Development in the Tropics and Subtropics. Amsterdam:Elsevier . INGRAM, M.J., FARMER, G. and WIGLEY, T.M.L. (1981)Pas t climates and their impact on man: areview . In: Climate and History (T.M.L. Wigley, M.J. Ingram and G.Farmer ,eds) ,pp .3-50 . Cambridge University Press. ISRAEL METEOROLOGICAL SERVICE (1967) Climatological Standard Normals of Rainfall, 1931-1960.Meteorologica l Notes Ser.A 21 ,Be t Dagan,Israel . ISSAR, A.S. and BRUINS,H.J . (1983)Specia l climatological conditions in the deserts of Sinai and the Negev during the latest Pleistocene. Palaeogeogra- phy, Palaeoclimatology,Palaeoecolog y 43:63-72. ISSAR, A., KARNIELI, A., BRUINS,H.J . and , I. (1984)Th e Quaternary geology and hydrology of Sede Zin, Negev, Israel. The Israel Journal of Earth Sciences33:34-42 . ISSAR, A.,KARNIELI ,A. ,BOWMAN ,D . and BRUINS,H.J . (1986)Residua l colluvio- aeolian aprons in the Negev highlands (Israel) -A paleo-climatic indicator. Accepted for publication in Palaeogeography, Palaeoclimatology, Palaeoecology. JARVIS, C.S. (1938)Deser t and Delta.London :Joh nMurray . JONGERIUS, A. and HEINTZBERGER, G. (1963)Th epreparatio n of mammoth-sized thin sections. Soil Survey Papers, nr. 1, Soil Survey Institute, Wageningen. JONGMANS, A.G. and MIEDEMA, R. (1985)Persona l communication. Agricultural University ofWageningen ,Departmen t of Soil Science and Geology. KAMPHORST, A. (1985) Personal communication. Agricultural University of Wageningen,Departmen t of Soils andFertilizers . KAMPHORST, A. and BOLT, G.H. (1978) Saline and sodic soils. In: Soil Chemistry. A. Basic Elements (G.H.Bol t and M.G.M. Bruggenwert,eds) ,pp . 171-191. Develpments in Soil Science 5A.Amsterdam :Elsevier . KARSCHON, R. (1984) Quseima: an oasis in the isthmic desert. Agricultural

209 Research Organization,Dept .o fForestry ,Leafle t No.78 .Ilanot ,Israel . KATZNELSON, Y. (1959)Th eclimat e of the Negev. Jerusalem: Meteorological Service Notes,Serie s 5(10). KEDAR, Y. (1957) Water and soil from the desert: some ancient agricultural achievements in the central Negev. TheGeographica l Journal123:179-187 . KEDAR, Y. (1967)Th eancien t agriculture in the Negev mountains. Jerusalem: Bialik Institute (in Hebrew). KEULEN, H. VAN (1979)Nitroge n nutrition,wate r use and plant production.In : Soils inMediterranea n Type Climates and their Yield Potential.Proceeding s 14th Colloq. International Potash Institute,Bern . KEULEN, H. VAN (1980)Modellin g grain production ofwhea t under conditions of limiting water supply: a case study. In: Proceedings 3rd International Wheat Conference (Madrid), pp. 414-419. (University of Nebraska-Lincoln, The Agricultural Experiment Station). KEULEN,H . VAN and DE WIT,C.T . (1975)Actua l and potential herbage production in arid regions. An annotated bibliography. Agricultural University of Wageningen, Dept. of Theoretical Production Ecology; and Institute for Biological and Chemical Research onFiel d Crops and Herbage (IBS). KHAZANOV, A.M. (1985) Nomads and the outside world. Cambridge: Cambridge University Press. KLEIN, C. (1961) On the fluctuations of the level of theDea d Sea since the beginning of the 19th century. Hydrological Paper No. 7. Hydrological Service,Ministr y of Agriculture, Jerusalem. KLEIN, C. (1981) The influence of rainfall over thecatchmen t area on the fluctuations of the level of the Dead Sea since the 12th century. Israel Meteorological Research Papers, Vol. 3. Naftali Rosenan Memorial Volume, pp. 29-58. Meteorological Service,Bet-Dagan . KLEIN, C. (1982)Morphologica l evidence of lake level changes,wester n shore of theDea d Sea. Israel Journal of Earth Sciences31:67-94 . KLONER, A. (1975)Ancien t agriculture atMamshi t and thedatin g of thewater - diversion systems in the Negev. Jerusalem: Israel Exploration Society, Eretz-Israel 12:167-170 (in Hebrew,wit h abstract in English). KRAEMER, C.J. (1958)Non-literar y Papyri. Excavations at Nessana,Volum e 3. Colt Archaeological Institute.Princeto n N.J.: Princeton University Press. KRENKEL, E. (1924) Der Syrische Bogen. Centralblat. Mineral. 9:274-281, 10:301-313. KUTIEL, H. (1978) The Distribution of Rain Intensities in Israel. M.Sc. Thesis, TheHebre w University of Jerusalem. LAMB, H.H. (1977)Climat e -Present ,Pas t and Future.Vol .2 .Climati c history and the future.London :Methuen . LAMB,H.H . (1982)Climate ,Histor y and the Modern World.London :Methuen . LAST, Y. (1978) ClassA Evaporatio n Pan data in Israel. Tahal, Tel-Aviv, Hydrology Department (in Hebrew). LAVEE, H. (1973) Relationship between the surface properties of the debris mantle and runoff yield in an extreme arid environment. M.Sc. Thesis,Th e Hebrew University of Jerusalem,Departmen t ofGeograph y (in Hebrew). LEE, Ch.A. VAN DER (1978)Grondslage n en aspecten van debewarin g van granen, peulvruchten, wortels en knollen in de tropen. Agricultural University of Wageningen,Departmen t of Tropical Crop Production. LE H0UER0U, H.N. (1985)Th e impact of climate on pastoralism. In: Climate Impact Assessment (R.W. Kates,J.H . Ausubel and M.Berberian ,eds) .Scien ­ tific Committee on Problems of theEnvironmen t (SCOPE), John Wiley k Sons. LE H0UER0U, H.N. andLUNDH0LM , B. (1976) Complementary activities for the improvement of the economy and the environment inmargina l . In: Can desert encroachment be stopped? Astud y with emphasis on Africa. Ecological Bulletin 24 (A.Rapp ,H.N . Le Houerou and B.Lundholm ,eds) ,pp . 217-229. Stockholm:NFR .

210 LEVY, T.E. (1983) The emergence of specialized pastoralism in the southern Levant.Worl d Archaeology 15:15-36. MALEY, J. (1973)Mecanism e des changementsclimatique s aux basses latitudes. Palaeogeography,Palaeoclimatology ,Palaeoecolog y 14:193-227. MARX,E . (1967)Bedoui n of the Negev.Manchester :Mancheste r University Press. MARX,E . (1977)Th e tribe as auni t of subsistence:Nomadi c pastoralism in the .America n Anthropologist 79:343-363. MARX, E. (1982)Ne w trends in the study ofpastora l nomads. Nomadic peoples 10:65-68. MAYERSON, P. (1955) Arid zone farming inAntiquity : A study of ancient agricultural and related hydrological practices in Southern Palestine. Ph.D. Thesis,Ne w York University. MAYERSON, P. (1960a) The ancient agricultural regime of Nessana and the Central Negeb.Col t Archaeological Institute.London : TheBritis h School of Archaeology in Jerusalem. MAYERSON, P. (1960b)Th e ancient agricultural remains of the central Negeb: Methodology and dating criteria.Bulleti n of the American Schools ofOrien ­ tal Research (BASOR)160:27-37 . MAYERSON, P. (1963) Thedeser t of southern Palestine according to Byzantine sources. Proceedings of the American Philosophical Society 107(2):160-172. MAYERSON, P. (1964) The first Muslim attacks on southern Palestine. Transactions and Proceedings of the American Philological Association 95:155-199. MAYERSON, P. (1967) A note on demography and land use in the ancient Negeb. Bulletin of the American Schools of Oriental Research (BASOR)185:39-43 . MCCLURE,H.A . (1971)Th e Arabian Peninsula and Prehistoric Populations.Miami , Florida:Fiel d Research Projects,Coconu t Grove. MIEDEMA, R., J0NGMANS, A.G. and SLAGER, S. (1974) Micromorphological observations on Pyrite and its oxidation products in four Holocene alluvial soils in TheNetherlands . In: Soil Microscopy (G.K. Rutherford, ed),pp . 772-794. Ontario:Th eLimeston e Press. M00K, W.G. (1983)14 Ccalibratio n curves depending on sample time-width. In: 14C and Archaeology (W.G.Moo k and H.T.Waterbolk , co-eds), PACT 8:517-525. M00K, W.G. (1985)Persona l communication. University ofGroningen , Isotope PhysicsLaboratory . MOOK, W.G. and STREURMAN,H.J . (1983)Physica l and chemical aspects ofradio ­ carbon dating. In: 14C and Archaeology (W.G.Moo k and H.T.Waterbolk ,co ­ eds), PACT 8:31-55. (Council ofEurope ,Strasbourg) . M00LEY, D.A. and PANT,G.B . (1981)Drought s in India over the last 200years , their socio-economic impacts and remedial measures for them. In: Climate and History (T.M.L. Wigley, M.J. Ingram and G.Farmer ,eds) ,pp .465-478 . Cambridge University Press. MULDERS, M.A. (1969)Th e arid soils of the Balikh Basin,Syria . Ph.D.Thesis , University ofUtrecht . MYERS, L.E. (1975) Water harvesting 2000 BC to 1974AD . Proceedings of the Water Harvesting Symposium, ,Arizona ,AR SW-22 ,USDA ,pp .1-7 . NATIV,R. , ISSAR,A .an d RUTLEDGE,J . (1983)Chemica l composition of rainwater and floodwaters in the Negev desert, Israel.J . of Hydrology 62:201-223. NEEV, D. and EMERY, K.O (1967)Th eDea d Sea. Depositional processes and environments of evaporites. Geological Survey of Israel (Jerusalem), Bulletin No.41 . NEEV/, D. and HALL, J.K. (1977)Climati c fluctuations during the Holocene as reflected by theDea d Sea levels. International Conference on Terminal ,Webe r State College,Ogden ,Utah . NEGEV, A. (1981)Le s Nabateans au Negev. Le Christianisme au Negev. La vie economique et sociale a l'epoque Byzantine.L e Monde de la Bible 18:4-46. NEGEV, A. (1982a)Numismatic s and Nabataean chronology. Palestine Exploration

211 Quarterly 114:119-128. NEGEV, A. (1982b) Christen und Christentum in der WusteNegev . Antike Welt 13(2):2-33. NEWMAN, J.C. (1963)Waterspreadin g on marginal arable areas. Journal of the Soil Conservation Service of New South Wales19:49-58 . NEWMAN, J.C. (1966)Waterpondin g for soil conservation in arid areas in New South Wales. Journal of the Soil Conservation Service of New South Wales 22:2-11 NICHOLSON, S.E. and FLOHN, H. (1980) African environmental and climatic changes and the general inLat e Pleistocene and Holocene. Climatic Change 2:313-348. NOY-MEIR, I. and SELIGMAN,N.G . (1979)Managemen t ofsemi-ari d ecosystems in Israel. In: Management of Semi-arid Ecosystems (B.H.Walker,ed) ,pp .113 - 160. Amsterdam:Elsevier . OATES, J. (1976)Prehistor y in Northeastern Arabia.Antiguit y 50(197):20-31. ORNI, E. and EFRAT, E. (1971) . Jerusalem: Israel UniversitiesPress . PEARSE, C.K. (1971)Grazin g in theMiddl e East: Past, present and future. Journal of Range Management 24(1):13-16. PEARSON, G.W. et al. (1985)Hig h precision radiocarbon timescale calibration. 12th International Radiocarbon Conference,Trondheim ,Norway . PEARSON, G.W. (1986) High precision radiocarbon timescale calibration. Radiocarbon (in press). PENNING DE VRIES, F.W.T. (1983)Th e productivity of Sahelian -a summary report. Overseas Development Institute (ODI), Agricultural Admini­ stration Unit,Pastora l Network Paper 15b. PENNING DE VRIES, F.W.T. and DJITAYE, M.A., eds (1982)L a productivite des Palurages Sahelians. Wageningen: Centre for Agricultural Publishing and Documentation (Pudoc). PLINY (the Elder). Natural History, transl.H . Rackham,W.H.S . Jones andE.E . Eichholz.Loe b Classical Library.London :Willia m Heineman. PONS, L.J. (1964)A guantitativ e microscopical method ofpyrit e determination in soils. In: Proceedings Micromorphological Symposium, Arnhem, The Netherlands (A.Jongerius ,ed) ,pp .401-409 . P0RAT, Y. (1981) Personal communication. Department of Antiquities, Rockefeller Museum, Jerusalem. R0SENAN, N. (1970)Temperature . Atlaso f Israel,IV/1 . Jerusalem: Survey of Israel,Ministr y ofLabour ;Amsterdam : Elsevier. ROTHENBERG, B. (1972)Timna .London : Thames and Hudson. SCHWARCZ, H.P., GOLDBERG,P . and BLACKWELL,B . (1980)Uraniu m series dating of archaeological sites in Israel. Israel J. ofEart h Sciences 29:157-165. SHANAN,L . (1975)Rainfal l and runoff relationships in smallwatershed s in the Avdat region of the Negev Desert Highlands. Ph.D. Thesis, The Hebrew University of Jerusalem,Departmen t of Geography. SHANAN,L. , EVENARI,M . and TADM0R, N.H. (1967)Rainfal l patterns in the Negev desert: Israel Exploration Journal 17(3):163-185. SHANAN, L., EVENARI,M . and TADM0R, N.H. (1969)Ancien t technology and modern science applied todeser t agriculture.Endeavou r 28(104):68-72. SHANAN, L. and SCHICK,A.P . (1980)A hydrologica l model for the Negev Desert highlands: Effects of infiltration, runoff and ancient agriculture. Hydrological Science Bulletin 25(3):269-282. SHANAN,L. , TADM0R, N.H.an d EVENARI,M . (1961)Th e ancient desert agriculture of the Negev, VII. Exploitation of runoff from large watersheds. Israel Journal ofAgricultura l Research (Ktavim)11:9-31 . SHANAN, L., TADM0R,N.H .an d EVENARI,M . (1970)Runof f farming in the desert, III. Micro-catchments for the improvement of desert range.Agronom y Journal 62:695-699.

212 SHANAN,L . and TADMOR,N.H . (1976,1979 )Micro-catchmen t systems for arid zone development, A handbook for design and construction. Rehovot, Israel: Ministry of Agriculture,Centr e for International Agricultural Cooperation. SHARON, D. (1972) Thespottines s of rainfall in adeser t area. Journal of Hydrology 17:161-175. SLICHER VAN BATH,B.H . (1960)D e agrarische geschiedenis vanWest-Europ a (500- 1850). Utrecht:Aula . SLICHER VAN BATH,B.H . (1963a)D e oogstopbrengsten van verschillende gewassen, voornamelijk granen, in verhouding tot het zaaizaad, ca.810-1820 .A.A.G . Bijdragen 9:29-125. Wageningen: Agricultural University, Department of Agrarian History. SLICHER VAN BATH, B.H. (1963b)Yiel d ratios,810-1820 .A.A.G .Bijdrage n 10:1- 264. Wageningen:Agricultura l University,Departmen t of Agrarian History. SPERBER, D. (1978)Roma n Palestine 200-400 - TheLand . Crisis and change in agrarian society as reflected in rabbinic sources. Ramat-Gan: Bar-Ilan University. SPITZ, P. (1980)Drough t and self-provisioning. In:Climati c constraints and human activities (J. Ausubel and A.K. Biswas, eds),pp . 125-147.Oxford : PergamonPress . STAGER,L.E . (1976)Farmin g in the Judean desert during the Iron Age.Bulleti n of the American Schools of Oriental Research (BAS0R)221:145-158 . STANHILL, G. (1961) A comparison of methods of calculating potential evapotranspiration from climatic data. Israel Journal of Agricultural Research 11:159-171. STANHILL, G. (1978) TheFellah' s farm: an autarkic agro-ecosystem. Agro- Ecosystems 4:433-448. TADMOR, N.H.,EVENARI ,M. , SHANAN,L . and HILLEL,D . (1958)Th e ancient desert agriculture in the Negev, I. Gravel mounds and strips near Shivta. Agricultural Reserach Station Rehovot,Ktavi m Records 8:127-151. THEOPHRASTUS. Enquiry into Plants,transl .A.F .Hort .Loe b Classical Library. London:Willia m Heineman. TH0RNE, D.W. and PETERSON, H.B. (1964) Irrigated soils:Thei r fertility and management.London :Th eBlackisto n Company. TURKOWSKI, L. (1969) Peasant agriculture in the Judaean hills. Palestine Exploration Quarterly (PEQ)101:21-33 ,101-112 . UNESCO (1979)Ma p of theworl d distribution of arid regions. Paris: Unesco, MAB Technical Notes7 . USDA (1975)Soi l Taxonomy. U.S. Department of Agriculture Handbook No. 436. Washington D.C.: U.S.Governmen t Printing Office. VARR0. De Re Rustica, transl. W.D. Hooper and H.B. Ash. Loeb Classical Library.London :Willia m Heineman. VITA-FINZI, C. (1969) The Mediterranean Valleys. Geological changes in historical times.Cambridge :Cambridg e University Press. VRIES, H. DE (1958)Variatio n inconcentratio n of radiocarbon with time and location on earth.Proceeding s Koninklijke Nederlandse Academie voorWeten - schappen,Ser .B ,61:94-102 . WAGSTAFF, J.M. (1981)Burie d assumptions:som e problems in the interpretation of the "Younger Fill" raised by recent data from Greece. Journal of Archaeological Science 8:247-264. WAISEL, Y., P0LLAK,G . and COHEN,Y . (1978)Th e ecology of the vegetation in Eretz Israel.Dept .o fBotany ,Tel-Avi v University. WAISEL, Y. and LIPHSCHITZ (1968)Dendrochronologica l studies in Israel. II. Juniperus Phoenica of north and central Sinai.La-Yaara n 1:2-22 (inHebrew , with English summary,pp . 63-67). WATSON, A.M. (1983)Agricultura l innovation in the early Islamic world. The diffusion ofcrop s and farming techniques, 700-1100. Cambridge: Cambridge University Press.

213 WEITZ, R. and ROKACH, A. (1968)Agricultura l Development: Planning and Implementation (Israel case study), Dordrecht:Reidel . WESTPHAL, E. (1974) Pulses in Ethiopia, their taxonomy and agricultural significance.Wageningen : Pudoc. WESTPHAL, E. (1982)Tropica l pulses. Agricultural University ofWageningen , Department of Tropical Crop Production. WHITE, K.D. (1963)Wheat-farmin g in Roman Times.Antiquit y 37:207-212. WIT, C.T. DE (1958) Transpiration and Crop Yields. Agricultural Research Reports 646.Wageningen :Pudoc . WIEDER, M. (1977)Occurrenc e and Genesis ofCarbonat e Nodules in Soils.Ph.D . Thesis. The Hebrew University of Jerusalem,Departmen t of Geology. WIEDER, M. and YAALON,D.H . (1974)Effec t ofmatri x composition on carbonate nodule crystallization. Geoderma11:95-121 . W00LLEY, C.L. and LAWRENCE, T.E. (1914/15, 1936)Th eWildernes s of Zin. London: Palestine Exploration Fund Annual 3 (1914/15). New York: Charles Scribner's Sons (1936). WOUDE, A.M. VAN DER (1963)D e consumptie van graan,vlee s en boter inHollan d op het einde van de achttiende eeuw. A.A.G. Bijdragen 9:127-153. Wageningen:Agricultura l University,Departmen t of Agrarian History. YAALON,D.H . (1963)O n the origin and accumulation of salts in groundwater and in soils of Israel.Bulleti n Research Council of Israel 11G:105-131. YAALON, D.H. (1964) Airborne salts as an active agent in pedogenetic processes. Bucharest: Trans.8t h Int.Congres s of Soil Science 5:997-1000. YAALON,D.H . (1981)Environmenta l setting. In:Aridi c Soilso f Israel (J.Dan , R. Gerson, H. Koyumdjisky and D.H. Yaalon,eds) ,pp .3-16 . Bet Dagan:Th e Volcani Center,Specia l Publication No.190 . YAALON, D.H. and GAN0R, E. (1973)Th e influence of dust on soils during the Quaternary. Soil Science 116:146-155. YAALON,D.H . and DAN,J . (1974)Accumulatio n and distribution of loess-derived deposits in the semi-arid and desert fringe areaso f Israel.Zeitschrif t f. Geomorphologie N.F.20:91-105 . YAIR, A. (1981)Th eSd e Boqer experiment site. In:Aridi c Soils of Israel (J. Dan, R. Gerson, H. Koyumdjisky and D.H. Yaalon,eds) ,pp .239-254 .Be t Dagan: The Volcani Center,Specia l Publication No.190 . YAIR,A . (1983)Hillslop e hydrology water harvesting and areal distribution of some ancient agricultural systems in the northern Negev desert. Journal of Arid Environments 6:283-301. YAIR, A. and DANIN,A . (1980)Spatia l variations in vegetation as related to the soil moisture regime over an arid limestone hillside, northern Negev, Israel. Oecologia47(l):83-88 . YAIR, A. and KLEIN,M . (1973)Th e influence of surface properties on flowan d erosion processes on debris-covered slopes in an arid area.Caten a 1:1-18. YAIR, A. and LAVEE,H . (1976)Runof f generative process and runoff yields from arid talusmantle d slopes.Eart h Surface Processes1(3):235-247 . YAIR, A., SHARON, D. andLAVEE ,H . (1978)A n instrumented watershed for the study of partial area contribution of runoff in the arid zone. Zeitschrift fur Geomorphologie N.F.29:71-82 . ZILBERMAN, E. (1981) The geology of the central Sinai-Negev shear Zone, central Negev. Part A, the Saad-Nafha lineament. Geological Survey of Israel, ReportHydro/1/81 . Z0HARY, D. (1973) The origin of cultivated cereals and pulses in the Near East. In: Chromosomes Today, Vol.4 (J.Wahrma n and K.R. Lewis,eds) ,pp . 307-320. New York:Joh nWiley . Z0HARY, D. and H0PF, M. (1973)Domesticatio n ofpulse s in the . Science182:887-894 . Z0HARY, M. (1966)Flor a Palaestina, Part I. Jerusalem: The Israel Academy of Sciences and Humanities.

214 SAMENVATTING

WOESTIJN MILIEU EN LANDBOUW IN DE CENTRALE NEGEV EN KADESH-BARNEA GEDURENDE DE HISTORIE

Ongeveer een derde deel van het oppervlak der aarde bestaat uit droge gebieden,di emeesta lworde nonderverdeel d inhyper-aride ,arid ee n semi-aride streken. Er bestaat eengroo t verschil inpotentiee l land-gebruik tussen deze drie droge zones, uitgaande vanhe t gebruik van locale regenval. In Israel ligt de semi-aride zone, waar landbouw m.b.v. directe regenval ind emeest e jaren mogelijk is, nette nnoorde n van deNege v woestijn. Degren stusse n de semi-aride en aridezon eval t hier samen met de30 0m m regenval isohyet. De noordelijke Negev en het tamelijk hoog gelegencentral e deel van de Negev liggen grotendeelsi nd earid e zone. De hyper-aride zuidelijkee n oostelijke Negev hebben eengemiddeld e jaarlijkse regenval vanminde r dan 60mm .Landbou w en veeteelt gebaseerd op locale regenval is ind ehyper-arid e zone nietmoge ­ lijk,behalv e inoases . De aride zone ismeesta l geschikt voor extensieve veehouderij, maar is te droog voornormal eregenafhankelijk e landbouw.Runof f landbouw,eveneen sgeba ­ seerd op locale regenval, isechte rwe lmogelij k ind earid e zone,indie nhe t milieu zich hiervoor leent:D egeomorfologische ,lithologisch e enbodemkundig e eigenschappen van het landschap, alsmedehe t regenval regime, moeten zodanig zijn dat regenwater relatief snelgaa t afstromen envervolgen s kan worden opgeslagen in hetbodemprofie l van de lageredele n van het landschap. Een dergelijk milieu isaanwezi g inhe t "RunoffLandbou w District"va n de centrale Negev, waar zich zeer veelcultuur-technisch e restanten van runoff landbouw bevinden,dateren d uithe tverleden . Runoff landbouw ishierbi jgedefinieer d als landbouw indrog e gebieden met behulp van afstromend regenwater (runoff)i nee nnatuurlij k of geprepareerd drainage gebied,o fvanui t een wadi. Op hydro-geomorfologische gronden worden vijfvorme n van runoff landbouw onderscheiden: 1)Micro-vanggebie d systemen 2)Geterrasseerd ewad i systemen 3)Helling-aquaduc t systemen A)Lima n systemen 5)Wadi-aftakkingskanaa l systemen

215 Het gebruik van runoff landbouw inne t aridedee l van de Negev woestijn gaat mogelijk terug naar het Chalcolithicum, zo'n550 0jaa r geleden. In de Israelitische periode of IJzer tijd (circa 1000voo rChristus ) werden een aantal runoff landbouw nederzettingen gevestigd in de centrale Negev. De archeologische datering van de velecultuur-technisch e overblijfselen van runoff landbouw die ind eNege v worden gevonden isechte r een moeilijke en complexe zaak, waarno g niet zovee l systematisch onderzoek aan is verricht. Een duidelijke chrono-stratigrafie ontbreekt in demeest egevallen . In een wadi-terras teHorva t Haluqim, aangelegd voor runoff landbouw ind e Israelitische periode ofno g daarvoor, werd microscopisch de aanwezigheid vastgesteld van ijzer-oxide concentraties in een ongeveer 3000 jaar oude bodem-horizont. Dit isee n duidelijk bewijs vanee nperiodie k nate ngeredu - ceerd bodem milieu inhe t verleden. Het practiseren van runoff landbouw in de oudheid,waarbi jhe t afstromend regenwater in de betreffende wadi werdvastge - houden door een reeks van lagekeer-damme n en aldus inhe t bodemprofiel werd opgeslagen, moet hiervoor verantwoordelijk zijn geweest. Dergelijke bodem- vorming indi t droge aride gebied isnormaa l nietmogelijk . Gedurende de historie zijn in het heuvelland vand ecentral e Negev, in een gebied van ongeveer 200.000hectare , zo'n A000hectar ecultiveerbar e grond in de wadis en valleien geschikt gemaakt voor runoff landbouw. Hiertoe werden duizenden dammen gebouwd in dewadis . Bovendien werden op vele hellingen eenvoudige doch functioneleaquaducte n aangelegd om meerafstromen d regenwater te kunnen verzamelen,hetgee n blijk geeft van een zeer goed ontwikkeld geomor- fologisch-hydrologisch begrip van het landschap ind e oudheid. Ook werden kanalen gegraven die de runoffwaterstrome n ind ewadi s ten dele konden aftakken en geleiden naar cultiveerbare velden. Tijdens de Byzantijnse periode, van de 5e-7eeeu w naChristus ,bereikt e de runoff landbouw in decentral e Negev woestijn zijn hoogtepunt. De gemiddelde tarwee n gerst oogsten zullen waarschijnlijk niet hoger zijn geweest dan circa 600 kg/hectare. Op grond van schattingen en berekeningen lijkt het erop dat minder dan een vierde deel der bevolking in deByzantijns e tijd kon worden gevoed met locaalverbouwd egrane n enpeulvruchten . Ditbevestig t de indruk dat het verder incultuu r brengen van de aride centraleNege v (grenzend aan de hyper-aride zone)d.m.v .runof f landbouw werd gestimuleerd en mogelijk gemaakt door de Byzantijnse autoriteiten. In dit verband moet worden gedacht aan het strategische voordeel van een sedentaire bevolking in dit zeer droge gebied ter verdediging van dezuid-oos t grens van het Byzantijnse rijk. Agrarische zelfvoorziening gebaseerd op runoff landbouw ismisschie n wel

216 mogelijk geweest met veel lagere bevolkingsaantallen invoorgaand e en latere perioden. Dit aspect vereist echter meer onderzoek. Interne vcedsel reserves moeten een integraal onderdeel hebben gevormd van runoff landbouw systemen ter overbrugging van deonvermijdelijk e jaren van extreme droogte enmisoogsten . Er lijkt geen relatie te bestaan tussen variaties vanhe t klimaat in het verleden en de geschiedenis van sedentaire runoff landbouw in de centrale Negev. Inee n deelva n de eerste eeuw voor Christuse n in deperiod e van circa 1200-1700A.D .wa s degemiddeld e regenval relatief hoger indi t gebied,e n het milieu derhalve beter geschikt voor runoff landbouw, dat echter in die eeuwen niet werd bedreven door een sedentaire bevolking. Demogelijk e activiteiten van semi-nomadischeBedoeine n op het gebied van runoff landbouw is historisch gezien niet te achterhalen voor de genoemde tijdvakken.

Opmerkelijke geomorfologisch-geologische veranderingen hebben zich in de laatste twee millennia voorgedaan in de vallei-oase vanKadesh-Barnea , gelegen in denoord-oostelijk eSinai .Sedimentair e ophoging (aggradatie) in deKadesh - Barnea vallei vond plaats in deRomeins e tijd, en vooral ind e periode van 1200-1700A.D . Tijdensd e aggradatie in dit laatste tijdvak werd het wadi-bed met circa A meter opgehoogd. Het permanentebeekj e inee n deel van de wadi, gevoed door een vand egrootst e bronnen van deNegev-Sina iwoestijn , stroomde hierdoor ook op eenhoge r niveau. Deze landschappelijke interpretatie werd overtuigend gestaafd door demicroscopisch e ontdekking vanauthigen e pyriet in het bovenste deel vand e aggradatie sedimenten. Na 1700heef t de wadi zich opnieuw in zijn eigen afzettingen ingesneden en ishe t niveau van het wadi bed in korte tijd met zo'n4 mete r verlaagd. Deze dramatische veranderingen ind e vallei van Ein elQudeira t (Kadesh- Barnea) vallen chronologisch samenme t bepaalde paleo-klimatologische perio­ den: (a)sedimentair e ophoging ind e vallei vond plaats tijdens een relatief nat klimaat; (b) insnijding en erosie viel samenme t een relatief droog klimaat. De geomorfologisch-geologische evolutie vand eKadesh-Barne a vallei inhe t Laat-Holoceen kan ongeveer als volgt worden samengevat:

1500- 100B.C . Dynamisch evenwicht 100B.C. -40 0A.D . Enige aggradatie en dynamisch evenwicht 400 -70 0A.D . Dynamisch evenwicht 700-120 0A.D . Insnijding en dynamisch evenwicht 1200-170 0A.D . Omvangrijke aggradatie 1700-198 6 A.D. Snelle insnijding en erosie

217 Irrigatie landbouw ind e vallei-oase van Kadesh-Barnea ofEi n el Qudeirat werd kennelijk reedsbedreve n in deBronstij d omstreeks 1600 voor Christus. Deze conclusie isgebaseer d op grond van deC-1 4 datering van een aquaduct restant. De overblijfselen van een dam met aquaductaa nhe t begin van de vallei-oase dateren volgensd e C-14 methode uit de7 eeeu wA.D . De aanwezig- heid van authigene pyriet in demorte l van dit aquaduct, dat zich bovenop de dam bevindt, isovertuigen d bewijs van natte gereduceerde omstandigheden in het verleden tijdens het functioneren van ditLaat-Byzantijn s - Vroeg-Arabisch irrigatie systeem. Het maximale aantal mensen dat in vroegere tijden konworde n gevoed in de vallei-oase van Kadesh-Barnea, d.m.v. irrigatie landbouw met water van de bron, bedroeg hooguit 300-400personen . Inhe t hele gebied van Quseimae nEi n el Qudeirat, door velen beschouwd alsd emees t waarschijnlijke streek van het Bijbelse Kadesh-Barnea, kunnen in deoudhei d nietmee r dan zo'n 1000 mensen geleefd hebben in systemen van agrarische zelfvoorziening. Aangezien de menigtede r Israelieten, onder aanvoering van Mozes, zeer zeker uit een veel grotere bevolking bestond tijdens het verblijf in de grotendeels hyper-aride Sinai woestijn, na deExodu s uitEgypte ,zo uhe t volk zonder de unieke voedsel-voorziening van manna zeker zijn omgekomen. Noch locale landbouwkundige productie, noch extensieve veehouderij, nochvoedsel - importen van devijandig e volken in de omgeving bieden enig wetenschappelijk alternatief in dit opzicht. Deze kwestie raakt de existentiele kern van de geschiedenis vanhe t volk Israel in diekritiek eperiod eva nhaa rbestaan . Nomadisch pastoralisme isee n gespecialiseerde vorm vanee n voedselprodu - cerende economie, die door z'n eenzijdigheid bijna niet tot autarkie kan leiden en in sterkemat e afhankelijk isva n de niet-nomadische sedentaire buitenwereld. Nomadisch pastoralisme leidt bijee ngelijkblijven d territorium tot stagnatie, omdat een wezenlijke toename vand e productie in een puur nomadische maatschappijvrijwe l onmogelijk is. Er zijn archeologische aanwij- zingen dat runoff landbouw systemen in de aride Negev werden gecombineerd met veehouderij. Met betrekking tot zelfvoorzienende voedsel productie gebaseerd op locale regenval inarid e streken verdient het aanbeveling de levensvatbaar- heid te onderzoeken van mogelijke combinaties van veehouderijme t de enige vorm van regenafhankelijke landbouw geschikt voor dearid e zone sensostricto : runoff landbouw. Voedselreserves voor jaren van extreme droogte zijn echter onmisbaar in dergelijke systemen.

218 CURRICULUM VITAE

The author was born in 1948 in Zeist,Th e Netherlands. In 1966h e completed secondary school (HBS-b)a t the Openbaar Lyceum "Schoonoord" in Zeist and began his studies at the Agricultural University ofWageningen . The author first studied landscape architecture, and later switched to soil science and geology, in which field hecontinue d at The Hebrew University of Jerusalem. In 1975h e received theDicker-Shrag a Award from theGeologica l Society of Israel and the Department ofGeolog y of The Hebrew University for his studies on the desert loess of the Negev.H e graduated cum laude in1978 . Since 1976 heha s been teaching soil science at the Ben-Gurion University of theNegev , and also worked for brief periods at TheHebre w University of Jerusalem and at Haifa University. In 1979-1980 hewa s a research fellow at the Department ofSoi l Science and Geology at theAgricultura l University of Wageningen, and investigated the paleomagnetic stratigraphy of the Maas terraces. From 1981-1984h eworke d for the Archaeological Survey of Israel to study ancient man-land relationships in the Negev desert and at Kadesh-Barnea. Since 1982h e has been connected with the Jacob Blaustein Institute for Desert Research of the Ben-Gurion University of theNegev . In 1984 he was co- organizer ofth e Third International Course onRunof fFarmin g inAri d Regions, held at the Sede Boqer Campus,Israel .

219