lable at ScienceDirect

Journal of Archaeological Science 35 (2008) 3035–3042

Contents lists avai

Journal of Archaeological Science

journal homepage: ht tp: / /www.elsevier .com/locate/ jas

Hydrogeological insights in antiquity as indicated by Canaanite and Israelitewater systems

Ram Weinberger*, Amihai Sneh, Eyal ShalevGeological Survey of Israel, 30 Malkhe Israel Street, Jerusalem 95501, Israel

a r t i c l e i n f o

Article history:Received 13 February 2008Received in revised form 17 June 2008Accepted 29 June 2008

Keywords:Water systemIron-AgeDead Sea FaultNear EastTell

* Corresponding author.E-mail address: rami.weinberger@gsi.gov.il (R. We

0305-4403/$ – see front matter � 2008 Elsevier Ltd.doi:10.1016/j.jas.2008.06.024

a b s t r a c t

Hazor, one of the largest fortified city during the Israelite (Iron-Age) period, encompasses a giganticunderground water system within its perimeter, supplying water to thousands of its inhabitants. It isconsidered as the ultimate example that Iron-Age engineers had mastered the concept of regionalgroundwater table. However, evaluating the hydrogeological conditions and the degree of success infinding water in Hazor and other contemporary cities, and thereby assess to what extent this concept wasknown at that time has seldom been done. Resurveying the water system of Hazor indicates that itswater chamber was dug along a major strand of the Dead Sea Fault, a boundary between the Arabian andAfrican (Sinai) plates. Hydrological simulations have shown that water ascends into the water chamber,utilizing this strand. Hazor’s engineers initially planned to connect the city with the springs at the foot ofthe mound (tell) through a shaft and tunnel as has been done at other biblical cities, but adjusted theirplan when groundwater was encountered within the mound perimeter. This accidental success as well asthe failure to reach the aquifer in deep water wells dug in the contemporary cities of Lachish and BeerSheba, imply limited hydrogeological understanding at the beginning of the first millennium B.C.E.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction

Ancient cities that were built on hilltops for defensive advan-tages suffered from the disadvantage that their natural watersources (e.g. springs and wadi floods) were outside the settlement –a fatal weakness at a time of siege. As rain seldom falls in the NearEast between the months of April and October, a permanentconnection to the nearby water supply is a matter of survival. Thisrequired the inhabitants either to dig wells into the aquifer underthe settlement, or to establish an underground connection toa water source outside the city. Such water systems were a typicalelement in the defenses of Canaanite and Israelite cities of theBronze and Iron ages (ca 3300–586 B.C.E.).

Water systems have been found in many sites in Israel (Fig. 1;Inset A), each displaying different hydrogeological conditions. Inseveral of these sites, however, constructing water systemsrequired engineering knowledge only. In all water systems in theCity of David, in Jerusalem (Reich and Shukron, 2004; Shiloh, 1992;Gill, 1991; Frumkin et al., 2003; Frumkin and Shimron, 2006) and inMegiddo (Lamon, 1935), the engineering mission was to connectthe cities with outside springs via shafts and tunnels; hydro-geological considerations were hardly involved. In Gibeon

inberger).

All rights reserved.

(Pritchard, 1961), the tunnel dug at the bottom of the ‘‘pool’’ wasunequivocally directed to reach a spring, not the groundwater,which was encountered a few meters from the spring (Cole, 1980).In Gezer – where dating is controversial (Macalister, 1912; Yadin,1969; Dever, 1969) – the tunnel reached groundwater collected ina natural cave, which might has a connection outside the mound(Reich and Shukron, 2003) (Fig. 2b). In the Canaanite city of Arad(Amiran et al., 1985), the construction constitutes a well-like cisternand not water well; groundwater table here is tens of meters belowthe bottom of the cistern (Fig. 2e).

In other sites, though, hydrogeological considerations wereessential in the construction of the water systems. The purpose ofthis study is to analyze these conditions, evaluate the degree ofsuccess in finding water and thereby assess to what extent theconcept of local and regional groundwater table in the Near Easthad been mastered during the Bronze and Iron ages.

2. Water search along the costal plain and stream banks

In sites located close to the Mediterranean Sea shore, e.g. TelNami (Artzy, 1991) or Tel Gerisa (Tsuk and Herzog, 1992) (Fig. 2a),where the groundwater table is only several meters below thesurface and the host rock is highly permeable sandstone, the ancientengineers possessed the knowledge of how to reach the ground-water since the Neolithic time (Galili and Nir, 1993). The same holdsfor wells that were dug in or along stream banks where floods

Fig. 1. A geological map of Tel Hazor. t – Turonian (Upper Judea Group); sp – Senonian–Paleocene; e1 – Lower Eocene; e2 – Middle Eocene; Pbd – Plio-Pleistocene Dalton Basalt; Qzg –Plio-Pleistocene Hazor–Gadot Formation; Al – Alluvium; R – ruins and anthropogenic strata; HWBF – Hula Western Border Fault (concealed along the mound). Area L (water system) ismarked by an arrow within the 9th–8th century (stratum VIII-V) city walls (red lines) after Ben-Tor (1993). Inset A: location map of Hazor and other Bronze and Iron ages sites in Israel.Insets B and C: plan and section of the water system of Hazor. Numbers denote altitude in meters above m.s.l. after Shiloh (1992). Location of Fig. 3 is indicated.

R. Weinberger et al. / Journal of Archaeological Science 35 (2008) 3035–30423036

recharge the shallow gravel aquifers and water can infiltratethrough porous sandstone banks, e.g. in Tel Haror (Oren et al., 1991)(Fig. 2d).

The two water wells, dug in Lachish and Beer Sheba, aresuggestive of limited understanding of hydrogeological systemsmore complicated than those described above. At Lachish,a 44 m deep well was dug at the lowest point on the mound’ssurface (Tufnell, 1953; Ussishkin, 1982), presumably intended toreach the groundwater table in the gravel aquifer of the LachishCreek running along the foot of the mound. Some water wasindeed reported to be found at the base of the cleaned up well.However, the well traversed and bottomed in Oligocene lime-stone and Eocene barely permeable chalk with no connection tothe gravel aquifer and therefore yielded insignificant quantitiesof water (Fig. 2c). A similar scenario is found at Tel Beer Sheba(Herzog, 2002), where flash floods occur in the nearby BeerSheba Creek several times every winter. Apparently, the inten-tion of digging the 69 m deep well at this site was to reachgroundwater table in the wadi (creek) gravel bed, assuming that

it would be stretching laterally under the mound, but the wellpenetrated mostly through Maastrichtian scarcely permeablechalk (Fig. 2f) which could hardly let through significant quan-tities of water.

3. The water system of Tel Hazor

The underground water system of Hazor, northern Israel(Shiloh, 1992; Yadin, 1969; Garfinkel and Greenberg, 1997)(Fig. 1), is an eloquent example of Iron-Age system serving anurban community. In 2005, Tel Hazor and its water system weredeclared a UNESCO world heritage site. As it is a systemsuccessfully hewn down to the groundwater table, unconnectedto any outside spring, the following inevitable question has tobe answered: was this an expression of a profound hydro-geological insight or merely an accidental success? In order tofathom this dilemma, an inquiry into the archeological as wellas into the geological and hydrogeological factors had beencarried out.

160

200180

220240

160

200180

220240

Eocene chalk Alluvium

Paleocene marl

sprin

g (proje

cted)

water chamber

Tel Gezer

b-S-

Elevationabove m.s.l.

-N-Elevation

above m.s.l.

water table

100 m0

180

220200

240260

280

180

220200

240260

280

Holocene gravel

Tel Lachish

Eocene chalk

well

Oligocene limestone

Eocene chalkT.D. 44 m

water table

Lachish

Creek

c

-SW-Elevation

above m.s.l.

-NE-Elevation

above m.s.l.

100 m0

Tel Haror

70

11090

130150

70

110

90

130150

well

Plio-Pleistocene sandstone

water table

Gerar

Creek

-S-Elevation

above m.s.l.

-N-Elevation

above m.s.l.

d

100 m0

Tel Arad

500

540520

560580

-SW-Elevation

above m.s.l.

Eocene chalk

Iron Age citadelEarly Bronze Age city

cistern (“well”)

-NE-Elevation

above m.s.l.

500

540520

560580

T.D. ~20 m

cistern

e

100 m0

Tel Beer Sheba

240

280260

300

320

240

280260

300320

220 220

Senonian chalk

Neogeneconglomerate

Holocene gravel

-N-Elevation

above m.s.l.

-S-Elevation

above m.s.l.well

T.D. 69 m

water table

Beer Sheba

Creek

water system (projected)

f

100 m0

-30

10-10

3050

-30

10-10

3050

water table

Ayalon

Creek

well

Tel Gerisa

Pleistocene calcareous sandstone

Alluvium

a-W-

Elevationabove m.s.l.

-E-Elevation

above m.s.l.

m.s.l.

100 m0

T.D. 11.5

Fig. 2. Water systems in principal sites in Israel (from north to south): (a) Tell Gerisa’s well (Middle Bronze Age) (Tsuk and Herzog, 1992). The shallow regional water table is about6 m above m.s.l. (b) Tell Gezer’s water chamber, which is a natural cave (Macalister, 1912; Reich and Shukron, 2003) at the contact between the impermeable Paleocene marl and thefractured Eocene chalk. Several springs flow along this contact; the closest, Ein Shusha, is projected onto the section. (c) The well of Lachish dated to stratum IV (Iron-Age II) orearlier (Ussishkin, 1982). The bottom of the well reached the level of Lachish Creek at the foot of the mound, suggesting that the intention was to reach its gravel aquifer. (d) The wellof Tell Haror dug above the northern bank of Gerar Creek (Oren et al., 1991) in permeable calcareous sandstone and reached the groundwater table. (e) The well-like cistern of Aradis located at the center of the depression within the walled area of the Early Bronze II city (Amiran et al., 1985). It was hewn into impermeable Eocene chalks to a depth of about20 m. The groundwater table at this locality is tens of meters below this level and there are no gravel aquifers or springs around. (f) The well in Beer Sheba (see text for details).

Fig. 3. An east-dipping fault in the northern wall of the Hazor’s water chamber (Fault Ain Fig. 4), throwing a conglomerate bed against a set of thin chalk beds. Oblique striaeobserved along the plane of Fault A, together with the stratigraphy, indicate an obliqueleft-lateral and normal displacement along this fault.

R. Weinberger et al. / Journal of Archaeological Science 35 (2008) 3035–30423038

3.1. Description of the Hazor’s water system

The water system was discovered by Yadin’s 1968–1969 expe-dition (Yadin, 1969; Garfinkel and Greenberg, 1997). It is located inthe upper city, adjacent to and within the city walls in area L (Fig. 1).Across from the location at the southern foot of the mound in HazorCreek, an aerial distance of approximately 150 m and an altitude of195 m above m.s.l., natural springs abound to this very day. Thewater system includes four elements (Shiloh, 1992; Yadin, 1969;Garfinkel and Greenberg, 1997): entrance structure, vertical shaft,sloping tunnel, and water chamber (Fig. 1; insets B and C). Theentrance structure includes two sloping ramps, connectingthe occupation level of the Israelite city at an altitude of 228 mabove m.s.l. with the head of the stairway of the vertical shaft. Theshaft is of trapezoid cross section, roughly 16�13 m, narrowingslightly with depth. Its upper part cuts through archaeologicalremains, belonging to stratum X [10th century B.C.E.; (Yadin, 1969;Ben-Tor, 1993)]; its lower part is quarried in bedrock conglomerateand chalk. The shaft is 19 m deep, and required the removal ofapproximately 6000 tons of debris and rocks. Five flights of stairsdescend around the shaft’s walls, the lowest along the southernwall of the shaft, turning at altitude 202 m above m.s.l. into thesloping tunnel. The tunnel runs straight for 25 m west-southwest,descending about 10 m and terminates into a 5� 5� 5 m chamberwhose bottom is covered with fresh water, forming a shallow poolat 36 m below the level of the entrance structure (192 m abovem.s.l.). The system allowed simultaneous movement of watercarriers, people as well as beasts going down or up. The entirewater system lies within the perimeter of the mound, providing theinhabitants of Hazor with a convenient approach to water duringtimes of peace, and, more important, at times of siege. Constructionof the water system is coeval with stratum VIII, dated to the 9thcentury B.C.E. and went out of use with the destruction of Hazor bythe Assyrian conquest (stratum V) in 732 B.C.E. (Yadin, 1972;Ben-Tor, 1993). It might have been damaged already by the 8thcentury B.C.E. earthquake (Amos 1:1), documented by destructionin stratum VI (Yadin, 1972).

Yadin (1969, 1972) attributed considerable importance to thefact that the sloping tunnel runs west-southwest, while the Hazorsprings are located at the foot of Tel Hazor to the south: ‘‘Thedirection of the tunnel came as a surprise, since we expected to findit to the south, in the general direction of the springs. However, thedeliberate and planned position and direction of the tunnel indi-cates that the engineers possessed sound geological knowledge. Itis obvious that they anticipated encountering the water-level – thesame as that of the springs – even within the perimeter of themound’’ (Yadin, 1969).

The idea that the 9th century B.C.E. engineers planned toencounter the groundwater table within the perimeter of themound is quite striking, mainly in light of the present knowledge ofthe hydrogeology of Hazor and its vicinity; the regional aquifer inthis area is tens of meters below the spring’s altitude. The upper-most part of the water system is 10 m higher than the lowest pointwithin the 9th century fortifications of the city (Fig. 1). If the ancientengineers dug towards the aquifer, why did they not locate thevertical shaft at the lowest (space-free) point on the mound and bythat saved much work and time? Hence, the location of the shaftclose to the southern edge of the mound, at closest proximity to thesprings, was clearly intentional.

3.2. Geologic setting of Tel Hazor

Tel Hazor is situated on the western margin of the Hula Valley,a large meridional depression along the segmented Dead Sea Fault(DSF). The DSF is an intracontinental left-lateral transformaccommodating the relative motion between the African (Sinai)

and Arabian plates. It is commonly agreed that the Hula Valley isa rhomb-shaped graben, which subsided in stages during the last4 Ma (Garfunkel, 1981; Heimann, 1990). The Hula basin is boundedby two longitudinal north-trending en-echelon strands of the DSF,known as the Hula Eastern Border Fault and Hula Western BorderFaults (HWBF), which are strongly expressed in the present-daymorphology. Transverse faults mark the northern and southernedges of the basin (Heimann, 1990; Rybakov et al., 2003), but theirsurface expression is subdued and secondary (Sneh andWeinberger, 2003a). A NNW-trending through-going left-lateralstrike-slip fault developed diagonally across the basin during themid-Pleistocene (Schattner and Weinberger, 2008). Consequently,subsidence has been controlled by both the basin bordering faultsand the diagonal fault, while the vertical displacement across thetransverse faults is minor. The Galilee Mountains rise steeply westof the HWBF. The surface trace of the fault is usually masked byPleistocene to recent sediments.

Tel Hazor is a man-made occupational mound. Its principal areaoverlies a natural hill rising 25–35 m above its surroundings. Thenorthwestern edge of the hill is built of the Dalton Basalt, dated to1.7 Ma (Heimann, 1990). The rest of the hill consists of horizontalbeds of conglomerate and chalk of the Plio-Pleistocene Hazor–GadotFormation, which was deposited contemporaneous with or laterthan the Dalton Basalt. The conglomerate beds consist of limestone,dolomite and flint pebbles derived from Cretaceous to Eoceneformations, and Plio-Pleistocene basalt pebbles derived from theDalton Basalt. The lower part of the vertical shaft, sloping tunnel andwater chamber are cut into 30 m of the Hazor–Gadot Formation.

35°’3

5

33°05’

35°’3

0

33°00’

‘Enan groundwater basin

Kinneret groundwater basin

Teo groundwater basin

‘Enansprings

Tel Hazor

HW

BF

Hazorsprings

A

A‘

0 2 km

Fig. 5. Boundaries between Teo, Enan and Kinneret groundwater basins. Arrowsindicate the groundwater flow directions. Location of cross section A–A0 (Figs. 7 and 8)is indicated.

AA

BC

D

B

DE

western wall

northern wallsouthern wall

1 m

ceiling

conglomeratechalkfault, sense of motion

Legend

Fig. 4. A panel diagram of the water chamber (looking west) and its faulted walls.Faults marked by letters. Differences in conglomerate thickness are depositional.

R. Weinberger et al. / Journal of Archaeological Science 35 (2008) 3035–3042 3039

In the course of geological mapping and delineation of thedifferent segments of the DSF system in and around the Hula Valley(Sneh and Weinberger, 2003b, 2006), we noticed that the DaltonBasalt is faulted against the Hazor–Gadot Formation by the HBWFat the northern foot of Tel Hazor (Fig. 1). Another strand of the faultwas delineated at the southern foot of the mound at the Hazorsprings. The fault appears to cross the hill of Tel Hazor through areaL. Therefore, we resurveyed the shaft, the sloping tunnel and thewater chamber. The fault trace was not identified in strata of theshaft and sloping tunnel, which are almost undisturbed beds ofconglomerate and chalk. However, these beds are intensivelyfractured in the water chamber, forming a series of sub-verticalfaults (Figs. 3 and 4). The western wall of the water chamberappears to coincide with a fault face (Fig. 4; Fault D) and fault traceson the southern wall are apposed to faults on the northern wall. Theeasternmost fault (Figs. 3 and 4; Fault A) is the most prominent,throwing a relatively thick conglomerate bed against a set of thinchalk beds. Oblique striae observed along the plane of Fault A,together with the stratigraphy, indicate an oblique left-lateral andnormal displacement along this fault. The fault zone is character-ized by a smearing of clay and a strong foliation. The strikes of thefault strands are between 355� and 10�, similar to the overall strikeof the HWBF.

3.3. Hydrogeological setting of Tel Hazor

To evaluate the influence of the faulting on groundwater flowinto the water chamber, we checked it within the framework of the

-W- Elevation

above m.s.l.

0 250 m

Eocene limestone(Avedat Gr.)

Senonian chalk(Mt. Scopus Gr.)

Cen

oman

ian-

Turo

nian

limes

tone

& d

olos

tone

(Jud

ea G

r.)

Plio& ch

200

100

300

400

0

Faul

t

Fig. 6. An east–west geological cross section

regional hydrogeological trends (Fig. 5). This shows that manysprings are aligned along the HWBF, where the permeable Creta-ceous to Eocene rocks of the Galilee Mountains are faulted againstimpermeable Plio-Pleistocene lacustrine and alluvial sediments ofthe Hula Valley (Grossowicz, 1969; Michelson and Goldstoff, 1974).Among these are the copious ’Enan springs, as well as the springs ofTeo, Hur, and Gome. The now established continuation of this faultfurther to the south reveals that the Hazor springs are also alignedalong this fault. Intake area for these springs is the Galilee Moun-tains, whence groundwater flows eastward towards the Huladepression, emerging upward along the HWBF (Grossowicz, 1969;Michelson and Goldstoff, 1974).

A two-dimensional numerical simulation was made ofgroundwater flow along a WSW–ENE cross section through TelHazor and its fault zone. The cross section traces the boundarybetween the Kinneret and the ’Enan groundwater basins (Fig. 5)and hence is parallel to a flow line, justifying the two-dimensionalapproach. Along this boundary, groundwater flows through theJudea aquifer from the recharge area at upper Amud Creek east-northeastward and discharges along the HWBF (Fig. 6).

It was assumed that the present-day hydraulic head is roughlythe same as at the time of the waterworks construction during the9th century B.C.E. Paleoclimate studies in central Israel show thatthe annual precipitation at that time was similar (within �10%) to

-E- Elevation

above m.s.l.

Tel Hazor

200

100

300

400

0

-Pleistocene cgl.alk(Hazor-Gadot Fm.)

Senonian chalk

Faul

t Waterchamber

springs

Judea aquifer groundwater

T.D. 36m

through the water chamber of Tel Hazor.

Table 1Hydraulic properties of rocks and fault zones used in the hydrogeological simulations

Unit Lithology Horizontal Permeability (m2)a Vertical Permeability (m2)a Porosity (%)

Judea Group Limestone, dolostone 5� 10�13 5� 10�15 10Mt. Scopus Group Clay, marl 10�17 10�19 5Avedat Group Marl, limestone 3� 10�13 3� 10�15 10Hazor–Gadot Formation Clay, marl 10�17 10�19 5

Conglomerate 5� 10�13 5� 10�15 10Fault zone Gravel 10�12 10�12 10

a Hydrologic properties of Judea Group, which is the main aquifer where flow occurs, are based on pumping tests (Michelson and Goldstoff, 1974). Hydrologic properties ofthe rest of the units are based on their lithology (Freeze and Cherry, 1979).

R. Weinberger et al. / Journal of Archaeological Science 35 (2008) 3035–30423040

present (Bar-Matthews and Ayalon, 2004). Noticeably, present-daylow flux of water in Hazor spring is affected by the pumping of freshwater in boreholes.

We solved the flow equation derived from Darcy’s law and theconservation of fluid mass along a cross section starting with theGalilean groundwater divide between the Mediterranean Sea andthe Hula Valley at the west, and Tel Hazor at the east (seeAppendix). The hydrological parameters of all lithological units aregiven in Table 1. The top of the impermeable Cretaceous LowerJudea Group is assigned as the bottom boundary with no flow(Fig. 7). Two vertical boundaries are also assigned with no flowconditions: the western boundary, which is the Galilean waterdivide, and the impermeable lithological units east of the HWBF.The first 2 km west of the water divide are assigned a constanthydraulic head as measured in boreholes. No recharge is assignedto the rest of the top boundary but discharge is allowed. The results

0 5,000distance

0

500

1,000

-500

elevatio

n (m

, ab

ove m

.s.l)

Upper Judea Gr

Mt. Scopus Gr.

Avedat Gr.

Lower200

150

100

10,000 12,00011,000

250

conglomerate

AvedatGr.

Hazor-Gadot

Fm.

HW

BF

chalk

distance (m)

WSW

194264Isr. coord.

A

elevatio

n (m

, ab

ove m

.s.l)

Fig. 7. A finite element mesh that follows a geological cross section between the Galilean grEocene Avedat Group is down-faulted to the west of HWBF; throw direction has later been

show that the hydraulic head decreases from the western boundaryeastward to the point of issue at Tel Hazor (Fig. 8). The simulationsshow that groundwater ascends along the HWBF into the mound,as well as discharges in local springs (Fig. 8; inset).

4. Synthesis

Based on the delineated fault strands and on the hydro-geological analysis, we suggest an explanation for the constructionof the underground water system of Hazor. The water system andits vertical shaft were located in the southern part of the mounddue to its close proximity to the nearby Hazor springs. In this it issimilar to other water systems in the Iron-Age in Israel, which werelocated as close as possible to the nearby water source at the foot ofthe mound (Lamon, 1935; Pritchard, 1961; Herzog, 2002). At analtitude of about 202 m above m.s.l, the sloping tunnel was

10,000 (m)

. Hazor-Gadot Fm.

Judea Gr.

faul

t

HW

BF

0

500

1,000

-500

ENE

205270Isr. coord.

elevatio

n (m

, ab

ove m

.s.l)

A‘

oundwater divide in the west and Tel Hazor in the east. For location see Fig. 5. Note, thereversed as shown in Fig. 1 (see also Fig. 6).

5,000 10,000distance (m)

0

500

1,000

-500

0

200 440320260 380

1 m/day

200

150

100

10,000 12,00011,000

250

distance (m)

0.5 m/day

190 220205195 210200 215

hydraulic head (m, above m.s.l.)

Upper Judea Gr.

Mt.Scopus Gr.

Avedat Gr.

Hazor-Gadot Fm.

Lower Judea Gr.

faul

t

MSMS

A

0

500

1,000

-500

WSW

194264Isr. coord.

ENE

205270Isr. coord.

Hul

a W

este

rn B

orde

r Fau

lt

elevatio

n (m

, ab

ove m

.s.l)

elevatio

n (m

, ab

ove m

.s.l)

elevatio

n (m

, ab

ove m

.s.l)

A A‘

Fig. 8. Results of hydrological simulations showing decreasing of hydraulic heads from the water divide in the west to Tel Hazor, where the groundwater discharges. Inset: details ofwater flow in and around the mound. Groundwater ascends along HWBF and flow laterally along the permeable conglomerate beds.

R. Weinberger et al. / Journal of Archaeological Science 35 (2008) 3035–3042 3041

constructed in a west-southwest direction sub-parallel to the wallsof the city. This direction was apparently chosen to avoid diggingtoo close to the outer face of the mound, notwithstanding thegreater distance. A horizontal tunnel should have been designed toconnect the end of the sloping tunnel with the springs. However,groundwater was encountered when they reached the fault zone atapproximate 195 m above m.s.l. and such a tunnel became redun-dant. The water chamber was set up exactly along this zone,utilizing the westernmost fault plane as the edge of the chamber. Itis also possible that during progressive quarrying of the slopingtunnel, water began to drip from the bedrock as a result ofascending pressurized groundwater along the HWBF and subse-quent lateral flow along porous conglomerate beds (Fig. 6 and 8).

The original plan in Hazor was to reach the nearby springs, butby lucky coincidence, strands of the DSF brought water into thecity’s perimeter. This accidental success as well as the failure todraw significant quantities of water in deep water wells dug in thecontemporary cities of Lachish and Beer Sheba, imply limitedhydrogeological understanding at the beginning of the firstmillennium B.C.E. It is intriguing to know that the same DSF thatwrought havoc in Israelite cities, killing their inhabitants, includingthe 8th century B.C.E. inhabitants of Hazor (Ben-Tor, 1993; Yadin,1972), also brought life giving groundwater to the city, saving itfrom drought during man-made sieges and wars.

Acknowledgments

This study was supported by grant no. 2004232 from the UnitedStates-Israel Binational Science Foundation (BSF). We are grateful toItamar Perath for his thorough reading and critical editing of themanuscript, and to Amnon Bentor and Amihai Mazar for helpfuldiscussions.

Appendix

The flow equation derived from Darcy’s law and the conserva-tion of fluid mass were solved along a cross section starting withthe groundwater divide between the Mediterranean Sea and theHula Valley at the west, and Tel Hazor at the east (Fig. 5). The flowequation and the conservation of fluid mass is (Bear, 1972):

V$½KVh� ¼ Ssvhvt

(1)

where K is the hydraulic conductivity tensor (L T�1), h is thehydraulic head (L), and Ss is the specific storage coefficient (L�1). Eq.(1) is solved using CPFLOW, a Galerkin finite element techniquewith linear shape functions applied over triangular elements fora two-dimensional cross section (Raffensperger, 1996). The top

R. Weinberger et al. / Journal of Archaeological Science 35 (2008) 3035–30423042

boundary is a free surface that represents the groundwater table(pressure¼ 0) and its position is calculated at every time step(Neuman and Witherspoon, 1971). Because the groundwater tableposition is changing in a stationary Cartesian coordinate plane(Eulerian formulation), it is necessary to describe the position of thegroundwater table within the elements. Every element is numeri-cally integrated over seven Gauss points at which the pressure iscalculated to find the actual position of the groundwater table.

References

Amiran, R., Goethert, R., Ilan, O., Alon, D., 1985. The water-system at Arad.Qadmoniot XVIII, 15–17.

Artzy, M., 1991. The Bronze Age anchorage site of Tel Nami. Qadmoniot XXIV(93–94), 31–38 (in Hebrew).

Bar-Matthews, M., Ayalon, A., 2004. Speleothems as paleoclimate indicators, a casestudy from Soreq cave located in the Eastern Mediterranean region, Israel. In:Battarbee, R.W., Gasse, F., Stickley, C.E. (Eds.), Past Climate variability throughEurope and Africa. Kluwer Academic Publishers, Dordrecht, The Netherlands,pp. 363–391.

Bear, J., 1972. Dynamics of Fluids in Porous Media. Elsevier, Amsterdam.Ben-Tor, A., 1993. Hazor. In: Stern, E. (Ed.), The New Encyclopedia of Archaeological

Excavations in the Holy Land, II, pp. 604–605.Cole, D., 1980. How water tunnels worked. Biblical Archaeology Review VI, 8–29.Dever, W.G., 1969. The water systems at Hazor and Gezer. Biblical Archaeologist 32,

71–78.Freeze, R.A., Cherry, J.A., 1979. Groundwater. Prentice-Hall, Englewood Cliffs, N.J.Frumkin, A., Shimron, A., Rosenbaum, J., 2003. Radiometric dating of the Siloam

Tunnel, Jerusalem. Nature 425, 169–171.Frumkin, A., Shimron, A., 2006. Tunnel engineering in the Iron Age: geo-

archaeology of the Siloam Tunnel, Jerusalem. Journal of Archaeological Science33, 227–237.

Galili, E., Nir, Y., 1993. The submerged pre pottery Neolithic water well of Atlit Yam,northern Israel, and its paleoenvironmental implications. The Holocene 3,265–270.

Garfinkel, Y., Greenberg, R., 1997. Area L. In: Ben-Tor, A., Bonfil, R. (Eds.), Hazor V, anAccount of the Fifth Season of Excavations, 1968, pp. 177–294.

Garfunkel, Z., 1981. Internal structure of the Dead Sea leaky transform (Rift) inrelation to plate kinematics. Tectonophysics 80, 81–108.

Gill, D., 1991. Subterranean waterworks of biblical Jerusalem: adaptation of a karstsystem. Science 254, 1467–1471.

Grossowicz, Y., 1969. The Hydrogeology of the Cenomanian – Turonian Aquifer ofthe Naftali Mountains. Diploma thesis, Hebrew University, Jerusalem, 20 pp., (inHebrew).

Heimann, A., 1990. The Development of the Dead Sea Rift and its Margins inNorthern Israel During the Pliocene and the Pleistocene. Geological Survey ofIsrael Report GSI/28/90, 83 pp., (in Hebrew; English abstract).

Herzog, Z., 2002. Water supply at Tel Beer Sheba in the 1st millennium B.C.E. In:Eran, Y. (Ed.), Cura Aquarum of Israel in Memoriam. Proceedings of the 11thInternational Water Conference on the History of Water Management andHydraulic Engineering in the Mediterranean Region, pp. 15–22.

Lamon, R.S., 1935. The Megiddo Water System, Chicago.Macalister, R.A.S., 1912. The Excavations of Gezer I–III, London.Michelson, H., Goldstoff, Y., 1974. The Geohydrology of the Cenomanian aquifer in

the Central and Upper Galilee. Tahal, Tel Aviv, 42 p. (in Hebrew).Neuman, S.P., Witherspoon, P.A., 1971. Analysis of non-steady flow with a free

surface using the finite element method. Water Resource Research 7,611–623.

Oren, E., Yekutieli, Y., Nahshoni, P., Feinstein, R., 1991. Tell Haror after six seasons.Qadmoniot XXIV (93–94), 2–19 (in Hebrew).

Pritchard, J.B., 1961. The Water System of Gibeon, Philadelphia.Raffensperger, J.P., 1996. Numerical simulation of sedimentary basin-scale hydro-

chemical processes. Advances in Porous Media 3, 185–305.Reich, R., Shukron, E., 2003. Notes on the Gezer Water System. Palestine Exploration

Quarterly 135, 22–29.Reich, R., Shukron, E., 2004. The history of the Gihon spring in Jerusalem. Levant 36,

211–224.Rybakov, M., Fleischer, L., ten Brink, U.S., 2003. The Hula Valley subsurface

structure inferred from gravity data. Israel Journal of Earth Science 52,113–122.

Schattner, U., Weinberger, R., 2008. A mid-Pleistocene deformation transition inthe Hula basin, northern Israel: implications for the tectonic evolution of theDead Sea Fault. Geochemistry Geophysics Geosystems 9, Q07009, doi:1029/2007GC001937.

Shiloh, Y., 1992. Underground water systems in the Iron age. In: Kempinski, A.,Reich, R. (Eds.), The Architecture of Ancient Israel from the Prehistoric to thePersian Periods, pp. 275–293. Jerusalem.

Sneh, A., Weinberger, R., 2003a. Geology of the Metulla quadrangle, northern Israel:implications for the offset along the Dead Sea Rift. Israel Journal of EarthScience 52, 123–138.

Sneh, A., Weinberger, R., 2003b. Geological Map of Israel. Sheet 2-II Metulla, scale1:50,000. Geological Survey of Israel.

Sneh, A., Weinberger, R., 2006. Geological Map of Israel. Sheet 2-IV Rosh Pinna, scale1:50,000. Geological Survey of Israel.

Tsuk, Z., Herzog, Z.,1992. The water system of Tel Gerisa (Israel) and its contribution tothe dating of underground water systems. In: Geschichte der Wasserwirtschaftund des Wasserbaus im Mediterranen Raum, Braunschweig, pp. 333–356.

Tufnell, O., 1953. Lachish III (Tell Ed-Duweir), the Iron Age, London.Ussishkin, D., 1982. The Conquest of Lachish by Sennacherib. Tel Aviv University.Yadin, Y., 1969. The fifth season of excavations at Hazor, 1968–1969. Biblical

Archaeologist 32, 50–71.Yadin, Y., 1972. Hazor: the Head of all those Kingdoms, London.