Journal of Archaeological Science 35 (2008) 956e964http://www.elsevier.com/locate/jas

Neutron activation and petrographic analysis of selected Late Bronzeand Iron Age pottery from Tell es-Safi/Gath, Israel

David Ben-Shlomo a,*, Aren M. Maeir b, Hans Mommsen c

a Institute of Archaeology, Hebrew University, Mount Scopus, Jerusalem 91905, Israelb Institute of Archaeology, Martin (Szusz) Department of Land of Israel Studies and Archaeology, Bar-Ilan University, Ramat-Gan 52900, Israel

c Helmholtz-Institut fur Strahlen- und Kernphysik, Universitat Bonn, Nussalle 14e16, Bonn 53115, Germany

Received 2 April 2007; received in revised form 16 June 2007; accepted 27 June 2007

Abstract

Thirteen sherds from the site of Tell es-Safi/Gath (Central Israel) were analyzed by neutron activation; six of these were also analyzed by thinsection petrography. These include mostly Late Bronze II and early Iron Age II imports (and possible imports) from Greece or Cyprus, as well asa sherd with a Late Bronze Age Egyptian Hieratic inscription. The details of the results are reported and discussed with their possible archae-ological implications.� 2007 Elsevier Ltd. All rights reserved.

Keywords: Ceramics; Provenancing; Late Bronze Age; Iron Age; Tell es-Safi/Gath; Philistia; Israel; Greece; Cyprus; Egypt; Imports; Neutron activation analysis;

Petrography

1. Introduction and background

Tell es-Safi/Gath is situated in central Israel, approximatelyhalfway between Jerusalem and Ashkelon, on the border be-tween the southern coastal plain and the Judean foothills(Shephelah) (Fig. 1). The site was settled virtually continu-ously from the Chalcolithic period until modern times. TheLate Bronze Age (LBA) settlement is identified as CanaaniteGath/Ginti, known from the el-Amarna letters, while duringthe Iron Age IeII (ca. 1200e586 BCE), the site is the locationof Philistine Gath, well known from biblical and Neo-Assyriantexts. To date, eleven seasons of survey and excavation havebeen conducted. In Fields A and E (Fig. 2), a stratigraphic se-quence spanning the latter part of the LBA (ca. 13th cent.BCE) through to the Iron Age IIB (ca. late 8th cent. BCE)was excavated (Maeir, 2003a). Four of the LBA samples ana-lyzed by Neutron Activation Analysis (NAA) were found in

* Corresponding author.

E-mail address: davben@mscc.huji.ac.il (D. Ben-Shlomo).

0305-4403/$ - see front matter � 2007 Elsevier Ltd. All rights reserved.

doi:10.1016/j.jas.2007.06.020

Field E, Stratum E4b, dating to the end of the Late BronzeAge, ca. 1200 BCE. The architectural evidence of this well-defined stratum was found in a number of squares in Field E,including several units. Of particular importance is a large,well-made building, of possible public function, in whicha wide range of finds were discovered, including quite a fewof cultic-oriented types. All told, the associated finds supporta terminal Late Bronze Age II dating, including local Canaanitepottery types, various Mycenaean and Cypriot imports, charac-teristic figurines and a collection of Egyptiaca (glyptics and twoEgyptian inscriptions; Maeir et al., 2004; Wimmer and Maeir,2007). It is noteworthy that wherever this level was exposed,the evidence indicated that it had been destroyed in a fire.

In the main area of excavation, Area A, located on the east-ern slopes of the tell, Stratum A3 was uncovered to an extentof over 1200 sq m. This was an extraordinarily well-preserveddestruction layer, with a wide range of finds, including over500 complete pottery vessels on the floors. Most of the build-ings seem to have had some industrial/agricultural functionwith several agricultural installations (Maeir, 2003a). Thisstratum was dated to the late 9th century BCE, based on

Fig. 1. Map of the eastern Mediterranean with sites and regions mentioned in text, with inset enlarged map of the southern Levant with sites mentioned in the text.

957D. Ben-Shlomo et al. / Journal of Archaeological Science 35 (2008) 956e964

a typological study of the local and imported pottery (includ-ing Cypriote ‘Black on Red’ juglets), and radiocarbon dates(Sharon et al., 2007: 44, Table 8). Three of the Iron Age sam-ples analyzed here derive from Stratum A3.

This limited study is a provenience analysis of eight LB IIand five Iron Age pottery vessels from Tell es-Safi/Gath(Table 1). This analysis complements a larger chemical andpetrographic study of the Iron Age pottery (Ben-Shlomo,2006: 183e187) and a petrographic study of Late BronzeAge pottery from Tell es-Safi/Gath. The main questions ofthis study were: (1) what was the origin of the MycenaeanIIIB pottery and various Iron Age II vessels imported to thesite?; (2) What was the provenance of a sherd with a Hieraticinscription (Sample SF160; Fig. 3:1), which according to mac-roscopic and microscopic investigation was not made froma known Egyptian fabric (Maeir et al., 2004: 130)?; (3)Determining the provenance of several vessels which weresuspected to be imported, or yielded indefinite results accord-ing to petrography. One example is a heavily red slipped sherdof a closed vessel with incised decoration (Sample SF149).This vessel has no known parallels, although the incised dec-oration resembles a 13th/12th century vessel from a rock-cutchamber tomb from Beirut, identified as ‘Grey BurnishedWare’ (possibly a globular bowl; Badre et al., 1998: 76,Fig. 4).

Greek Iron Age imports at Iron Age sites in the SouthernLevant have drawn much attention, due to their potential toserve as chronological markers for both the Greek and Levan-tine sequences of the early 1st millennium BCE, as well as in-dicators for Greek contacts with the Levant during this period(see, e.g. Waldbaum, 1994; Fantalkin, 2001). Late Iron I/earlyIron II Greek imports are very rare in Palestine, and thus, the‘Proto-Geometric’ sherd (Sample SF105) was included in thisstudy, also as its typological identification was inconclusive: itwas either a Proto-geometric Ionian (East Greek) vessel, ora Proto-geometric/Sub-Mycenaean vessel from the mainland.

2. Analytical procedures

The NAA performed in Bonn is a modified version of theprocedure given by Perlman and Asaro (1969) and is describedin more detail in Mommsen et al. (1991). The thirteen sampleswere drilled from the sherds using a diamond point, andweighed to about 80 mg. The samples were irradiated at theresearch reactor at Geesthacht (GKSS). The concentration ofthirty elements was obtained by comparison of the Bonn stan-dard, calibrated against the well-known Berkeley Pottery Stan-dard. The assignment of individual samples to establishedchemical groups was made according to the statistical filteringmethod developed in Bonn (Beier and Mommsen, 1994;

Fig. 2. General plan of the excavations at Tell es-Safi/Gath.

958 D. Ben-Shlomo et al. / Journal of Archaeological Science 35 (2008) 956e964

Mommsen et al., 2002: 615e617), which uses a modified Ma-halanobis distance as a dissimilarity measure, and takes intoaccount both the different sizes of the individual elementalmeasuring errors and various dilution effects, which result inconstant shifts in elemental values. These dilution effects aredue to varying pottery making practices and/or to a naturalvariation of the raw materials. The high precision of the resultsas well as a data bank of over 7000 samples from the easternMediterranean, mainly from Greece and the Aegean region,

Table 1

List of samples from Tell es-Safi/Gath

Sample Fig. # NAA group TS group Description

SF161 (NAA10) Fig. 3:1 MYBE e Mycenaean IIIB s

jar spout

SF159 (NAA8) MBKR e Mycenaean IIIB k

with chariot whee

SF163 (NAA12) GR mainland e Mycenaean IIIB s

(krater?)

SF160 (NAA9) NegA e Sherd with Hierati

inscription

SF142 (NAA11) Fig. 3:2 NegA Brown b Bichrome krater

SF149 (NAA13) NegA Loess Incised and heavil

red-slipped sherd

SF158 (NAA6) PalJ e Phoenician style

‘fluted’ bowl

SF156 (NAA3) Fig. 3:3 PalJ ass. e Pre-LMLK Jar

SF152 (NAA5) Fig. 3:4 PalJ Loessþ Cal. ‘Syrian’ style Figu

SF32 (NAA1) Edom ass. Motza? ‘Ajrud’ type jar

SF104 (NAA2) EME-A Imported ‘White Painted’ sh

SF105 (NAA7) MYBE Imported ‘Proto-Geometric’

skyphos

SF157 (NAA4) Fig. 3:5 Single e ‘Black on Red’ ju

but including also about 50 samples from Cyprus, 460 fromPalestine and 120 from Egypt, enables us to confidently assignmost samples to chemically-homogeneous groups, that aregeographically provenanced according to earlier research byMommsen (e.g., Mommsen et al., 1988, 2002).

Thin-section petrographic analysis was carried on six of thesamples (several of the sherds were either too small to be sam-pled, or were made of very fine clay). The analysis was carriedin the Institute of Archaeology of the Hebrew University.

Basket Locus Stratum Context Period

tirrup 680011/1 68001 e Topsoil LBII

rater

l

580029 58010 E3 From an Iron I

Rubbish pit

LBII

herd 680063 68008 E4b Destruction Debris LBII

c 580115 58021 E4b Occupation Debris LBII

680015/1 49023 E4b Occupation Debris LBII

y 680104 68008 E4b Destruction Debris LBII

520020 44014 A3 Debris Iron IIA

230362 23021 A3 Destruction Debris Iron IIA

rine 360004 Survey e Found in section made

in Area E

LBA

220271 22021 A3 Destruction Debris Iron IIA

erd 230296/1 23028 e Topsoil Iron II

530208/1 53023 A4 Fill Iron IIA

glet 320204 32013 e Topsoil Iron IIA

Fig. 3. Selected line drawings of pottery from Tell es-Safi/Gath provenanced

by NAA. 1. Sherd with Egyptian Hieratic inscription (SF160); 2. LBA Bi-

chrome Krater (SF142); 3. Iron Age IIA ‘Pre-LMLK’ Jar (SF156); 4 ‘Syrian

Style’ figurine (SF152);. 5. Iron Age IIA ‘Black on Red’ Juglet (SF157).

Fig. 4. Top: Graphical comparison of chemical compositions of the group

MYBE and the sherd SF 105 (plotted are the differences of the concentration

values MYBE e SF 105 normalized by the average standard deviations. The

values of sherd SF 105 are multiplied first by the best relative fit factor with

respect to group MYBE of 1.01. The concentrations are statistically similar.

Bottom: same as top for group MYBE and sherd SF 159 multiplied by the

best relative fit factor with respect to group MYBE of 0.96 (sherd SF 159 is

assigned to a subgroup MBKR of MYBE with lower potassium (K) and rubid-

ium (Rb) and sometimes higher sodium (Na) concentrations).

959D. Ben-Shlomo et al. / Journal of Archaeological Science 35 (2008) 956e964

Samples were obtained by sectioning a pottery sherd to a thick-ness of 0.03 mm (30 microns) and slides were examined witha petrographic polarizing microscope (a Nikon Labophot-Poland Zeiss microscopes were used, with magnifications of�25e�400). Inclusions were identified on the basis of theirmorphological characteristics. Petrographic groups were deter-mined according to the various characteristics observed in theslides as matrix soil type, and the type, texture, and quantityof the inclusions (for detailed procedure see Ben-Shlomo,2006: 134e136).

3. Results

3.1. NAA (Table 2)

The Mycenaean IIIB stirrup jar (SF161) matches in compo-sition the MYBE group previously identified by Mommsen(below, Table 3: Col. 1; Mommsen et al., 1988). Pottery as-signed to this group is believed to have been manufacturedat the Mycenaean workshop at Berbati (Fig. 1), since kilnwasters from there have this profile (Mommsen et al., 2002:621), and probably also at other workshops at later periodsin the Argolid and the north-eastern Peloponnese. This groupnow includes 165 members found in the Argolid (Table 3: Col.1) and many more pieces exported from this region to otherGreek and more distant areas. Another fragment of a Myce-naean IIIB chariot krater (SF 159) belongs to group MBKR(Fig. 4: bottom), which is a variant of group MYBE with lowerpotassium and rubidium values than samples assigned toMYBE (Mountjoy and Mommsen, 2001: 125, Table 1: groupMBKR). The chemical data for the third Mycenaean IIIBsherd (SF163) does not fit statistically any of the well assignedpatterns in the Bonn databank. However, it is closest to groupMBKR and also is associated to patterns from the western andeastern Peloponnese and Attica. We therefore assume thissample is from the Greek mainland/Peloponnese, but are un-able to provide a more firm provenance at this time. The

Table 2

Tell es-Safi/Gath sherds

Sample Factor As Ba Ca % Ce Co Cr Cs Eu Fe % Ga

SF161-10 1.000 5.07 532.0 10.6 58.9 26.0 193.0 7.50 1.08 4.65 18.3

SF159-8 1.000 3.82 398.0 9.87 67.8 28.2 222.0 8.97 1.17 5.36 26.5

SF163-12 1.000 3.49 339.0 10.2 59.7 26.5 201.0 6.83 1.17 4.54 21.6

SF160-9 1.000 4.12 378.0 11.8 56.0 12.5 87.9 1.17 1.07 3.01 9.19

SF142-11 1.000 5.57 405.0 8.52 67.2 17.2 113.0 1.86 1.44 3.95 17.7

SF149-13 1.000 4.01 260.0 14.8 49.3 12.0 91.4 1.11 1.08 2.83 16.7

SF158-6 1.000 3.84 846.0 5.77 70.8 20.1 115.0 1.79 1.55 4.56 20.9

SF156-3 1.000 3.99 1089.0 7.42 67.6 21.3 122.0 1.43 1.42 4.29 23.0

SF152-5 1.000 6.03 654.0 15.0 66.4 17.7 108.0 2.55 1.42 4.27 21.7

SF32-1 1.000 5.55 610.0 4.25 50.8 15.1 108.0 5.30 1.11 4.04 23.9

SF104-2 1.000 8.51 696.0 11.9 38.1 29.7 359.0 4.17 0.97 5.81 18.0

SF105-7 1.000 4.97 469.0 10.3 62.3 25.3 215.0 7.74 1.11 5.05 22.1

SF157-4 1.000 4.59 313.0 2.50 84.3 23.9 90.8 4.32 1.44 4.15 19.2

ave.meas.error 0.089 14.0 0.16 0.82 0.12 0.88 0.078 0.022 0.016 2.1

in % 1.8 2.6 1.7 1.3 0.6 0.6 1.9 1.8 0.4 11.

Sample Factor Hf K % La Lu Na % Nd Ni Rb Sb Sc

SF161-10 1.000 3.68 2.49 29.0 0.41 0.56 25.9 276.0 130.0 0.44 18.7

SF159-8 1.000 3.90 1.81 32.7 0.45 1.29 28.1 217.0 95.5 0.62 21.5

SF163-12 1.000 3.95 2.30 30.2 0.42 0.67 26.4 298.0 114.0 0.54 17.4

SF160-9 1.000 8.34 1.54 25.8 0.33 0.46 24.1 278.0 33.9 0.34 10.3

SF142-11 1.000 12.6 1.60 32.3 0.49 0.65 30.9 220.0 51.8 0.49 13.8

SF149-13 1.000 9.98 1.44 23.7 0.39 0.47 21.3 68.9 34.7 0.33 9.50

SF158-6 1.000 10.9 1.12 32.7 0.51 0.85 32.1 110.0 52.8 0.68 15.0

SF156-3 1.000 10.3 1.33 32.4 0.50 0.43 30.5 174.0 44.6 0.48 14.6

SF152-5 1.000 8.44 1.16 32.2 0.47 0.47 31.3 207.0 47.8 0.49 14.1

SF32-1 1.000 4.47 4.84 23.1 0.38 0.20 23.2 118.0 110.0 0.30 17.0

SF104-2 1.000 3.03 1.87 18.6 0.44 1.05 18.6 275.0 71.3 0.87 25.2

SF105-7 1.000 3.74 2.73 30.9 0.44 0.70 26.6 268.0 132.0 0.58 20.0

SF157-4 1.000 4.06 1.38 34.1 0.40 0.34 32.1 222.0 72.6 0.61 15.1

ave.meas.error 0.066 0.025 0.16 0.016 0.001 0.78 33.0 2.0 0.057 0.021

in % 1.0 1.3 0.5 3.7 0.1 2.9 16.0 2.6 11.0 0.1

Sample Factor Sm Ta Tb Th Ti % U W Yb Zn Zr

SF161-10 1.000 4.54 0.71 0.64 10.0 0.45 2.21 2.30 2.72 112.0 51.1

SF159-8 1.000 5.02 0.85 0.72 11.4 0.23 2.30 1.77 2.88 110.0 50.6

SF163-12 1.000 4.75 0.72 0.69 9.63 0.26 2.11 1.94 2.84 113.0 92.4

SF160-9 1.000 4.14 1.09 0.67 6.78 0.27 1.62 0.93 2.52 79.6 276.0

SF142-11 1.000 5.60 1.30 0.84 8.77 0.38 3.35 1.28 3.53 85.6 396.0

SF149-13 1.000 3.85 0.91 0.69 5.99 0.33 2.28 1.07 2.72 77.9 300.0

SF158-6 1.000 5.92 1.33 0.87 8.71 0.55 2.65 1.44 3.26 74.8 348.0

SF156-3 1.000 5.56 1.26 0.84 7.99 0.66 2.07 1.64 3.38 110.0 348.0

SF152-5 1.000 5.63 1.25 0.78 7.61 0.53 2.43 1.46 3.24 86.0 291.0

SF32-1 1.000 4.38 0.81 0.68 7.55 0.45 1.87 1.40 2.40 54.1 114.0

SF104-2 1.000 3.45 0.56 0.53 5.86 0.44 2.64 1.68 2.44 100.0 36.4

SF105-7 1.000 4.88 0.74 0.64 10.6 0.43 2.54 2.69 2.78 113.0 60.7

SF157-4 1.000 5.77 1.14 0.77 10.5 0.36 1.61 2.21 2.62 98.3 143.0

ave.meas.error 0.011 0.029 0.043 0.056 0.062 0.085 0.14 0.048 2.1 27.0

in % 0.2 2.9 6.0 0.7 15.0 3.7 8.5 1.7 2.2 14.

Concentrations of elements C measured by NAA in mg/g (ppm), if not indicated otherwise, and average errors, also in percent of C.

960 D. Ben-Shlomo et al. / Journal of Archaeological Science 35 (2008) 956e964

Proto-geometric/sub-Mycenaean sherd (SF105; Table 3: Col.2; Fig. 4: top) also matched the MYBE group, and thus wasproduced in the area of the Argolid.

The sherd with the Hieratic inscription (Fig. 3:1; SF160;Table 3: Col. 4) is not very different in composition to a chem-ical pair, a Hittite Bulla found at Tel Aphek and a krater (N48)from the Kadesh-Barnea fortress of the seven-sixth centuryBC (Goren et al., 2007). A Canaanite Bichrome decoratedLBII krater (Fig. 3:2; SF142) and a red slipped and incised

sherd of an unidentified ware (SF149) also belonged to thesame chemical group titled now NegA (Table 3: Col. 3).This group was formed only tentatively, since it has an unusualhigh spread in the hafnium values, which suggests possiblesubgroups. Also, the individual best relative fit factors of these5 samples are very different (see Table 3) and point to claypastes with different diluting admixtures. The sherd with theHieratic inscription (Fig. 3:1; SF160) and the red slippedand incised sherd (SF149) have absolute concentration values

Table 3

Concentrations of elements measured by NAA: Group averages M or individual samples C in mg/g (ppm), if not indicated otherwise, and spreads s or errors d in

percent

MYBE SF105 NegA SF160 PALJ SF152 Edom EME-A

165 samples 1 sample 5 samples 1 sample 25 samples 1 sample 26 samples 39 samples

factor 1.00 factor 1.02 factor 1.00 factor 1.09 factor 1.00 factor 0.91 factor 1.00 factor 1.00

incl. SF 161 to match MYBE to match NegA incl.SF156, 158 to match PALJ incl. SF32 incl. SF104

Mþ/� s(%) Cþ/� d(%) Mþ/� s(%) Cþ/� d(%) Mþ/� s(%) Cþ/� d(%) Mþ/� s(%) Mþ/� s(%)

As 6.03 (50.0) 5.07 (2.0) 4.88 (39.0) 4.45 (1.8) 4.46 (15.0) 5.49 (1.5) 7.66 (50.0) 8.77 (19.0)

Ba 396.0 (23.0) 478.0 (3.0) 325.0 (18.0) 408.0 (3.1) 557.0 (32.0) 595.0 (2.2) 326.0 (73.0) 449.0 (25.0)

Ca % 9.72 (23.0) 10.5 (1.6) 9.99 (46.0) 12.8 (1.4) 7.22 (43.0) 13.6 (1.2) 3.40 (87.0) 9.95 (19.0)

Ce 62.9 (2.6) 63.5 (1.2) 56.7 (4.3) 60.4 (1.3) 64.2 (5.9) 60.4 (1.2) 55.2 (5.6) 35.4 (3.4)

Co 27.7 (4.7) 25.8 (0.5) 13.8 (3.7) 13.5 (0.7) 17.5 (4.5) 16.1 (0.6) 17.1 (15.0) 27.2 (5.0)

Cr 211.0 (7.6) 219.0 (0.5) 98.2 (4.8) 94.9 (0.7) 100.0 (4.1) 98.7 (0.7) 120.0 (17.0) 327.0 (26.0)

Cs 8.66 (8.1) 7.89 (1.2) 1.37 (9.9) 1.26 (5.0) 1.32 (17.0) 2.32 (2.7) 3.83 (24.0) 3.35 (9.9)

Eu 1.15 (4.2) 1.14 (2.0) 1.16 (4.4) 1.15 (1.9) 1.34 (3.2) 1.29 (1.6) 1.25 (7.7) 0.87 (3.9)

Fe % 5.14 (3.4) 5.15 (0.3) 3.14 (4.9) 3.25 (0.4) 3.82 (4.1) 3.89 (0.4) 4.34 (8.9) 5.18 (3.5)

Ga 21.0 (20.0) 22.6 (12.0) 13.1 (28.0) 9.93 (18.0) 14.7 (16.0) 19.7 (9.9) 35.6 (29.0) 16.3 (13.0)

Hf 3.60 (8.5) 3.82 (1.5) 11.4 (16.0) 9.01 (0.8) 10.9 (15.0) 7.68 (0.8) 4.91 (17.0) 2.86 (8.5)

K % 2.63 (8.1) 2.79 (1.1) 1.74 (18.0) 1.67 (1.3) 1.13 (19.0) 1.06 (1.9) 2.18 (18.0) 1.56 (13.0)

La 31.5 (2.2) 31.6 (0.5) 26.6 (3.9) 27.9 (0.5) 28.8 (2.9) 29.3 (0.5) 25.6 (6.1) 16.5 (3.8)

Lu 0.42 (5.7) 0.45 (4.1) 0.41 (9.8) 0.35 (3.7) 0.44 (5.0) 0.43 (3.6) 0.41 (6.4) 0.36 (6.0)

Na % 0.52 (25.0) 0.71 (0.1) 0.69 (46.0) 0.49 (0.2) 0.63 (32.0) 0.43 (0.1) 0.61 (89.0) 0.91 (15.0)

Nd 26.5 (5.4) 27.1 (3.1) 24.2 (10.0) 26.0 (2.8) 26.5 (3.6) 28.5 (2.7) 24.7 (6.5) 15.0 (10.0)

Ni 207.0 (11.0) 274.0 (13.0) 138.0 (75.0) 300.0 (10.0) 61.3 (37.0) 188.0 (15.0) 49.8 (25.0) 221.0 (18.0)

Rb 151.0 (5.7) 135.0 (1.9) 40.4 (12.0) 36.6 (4.3) 41.6 (11.0) 43.5 (3.6) 70.8 (21.0) 55.8 (9.9)

Sb 0.58 (15.0) 0.59 (11.0) 0.37 (22.0) 0.37 (14.0) 0.48 (20.0) 0.44 (11.0) 0.35 (27.0) 0.72 (14.0)

Sc 21.0 (3.3) 20.4 (0.1) 10.8 (5.3) 11.2 (0.2) 12.7 (4.5) 12.8 (0.1) 19.7 (13.0) 21.8 (4.9)

Sm 4.86 (4.3) 4.98 (0.2) 4.43 (7.3) 4.47 (0.2) 5.10 (2.8) 5.12 (0.2) 4.98 (6.6) 3.11 (3.9)

Ta 0.80 (6.2) 0.75 (3.7) 1.09 (6.1) 1.18 (2.5) 1.14 (5.4) 1.14 (2.4) 0.81 (20.0) 0.54 (5.4)

Tb 0.68 (6.8) 0.66 (6.9) 0.71 (6.9) 0.73 (5.6) 0.76 (4.6) 0.71 (5.3) 0.69 (9.3) 0.51 (11.0)

Th 10.9 (2.8) 10.8 (0.6) 7.25 (5.5) 7.32 (0.7) 7.34 (5.6) 6.92 (0.7) 7.30 (7.9) 5.59 (4.8)

Ti % 0.43 (22.0) 0.44 (15.0) 0.44 (44.0) 0.29 (23.0) 0.65 (12.0) 0.49 (11.0) 0.49 (9.1) 0.52 (22.0)

U 2.28 (6.5) 2.59 (3.5) 2.28 (22.0) 1.75 (4.7) 1.86 (23.0) 2.21 (3.3) 2.34 (30.0) 1.69 (16.0)

W 2.18 (13.0) 2.75 (6.1) 1.15 (12.0) 1.00 (13.0) 1.31 (18.0) 1.33 (9.4) 2.11 (19.0) 1.53 (19.0)

Yb 2.75 (3.2) 2.83 (1.8) 2.95 (4.3) 2.72 (1.8) 2.95 (4.0) 2.95 (1.5) 2.76 (8.5) 2.08 (4.4)

Zn 109.0 (8.5) 115.0 (2.1) 81.4 (7.7) 86.0 (2.2) 75.6 (32.0) 78.3 (2.3) 72.4 (39.0) 101.0 (18.0)

Zr 151.0 (25.0) 61.9 (44.0) 401.0 (27.0) 298.0 (7.9) 365.0 (23.0) 265.0 (8.9) 214.0 (20.0) 88.1 (36.0)

The averages M have been calculated after correcting the individual samples by a best relative fit with respect to M for dilution (fit factors for group NegA: Hittite

Bulla 1.06, N48 0.93, SF142 0.84, SF149 1.11, SF160 1.09, see text).

961D. Ben-Shlomo et al. / Journal of Archaeological Science 35 (2008) 956e964

comparable to the Hittite Bulla, whereas the two kraters(SF142 [Fig. 3:2] and N48) are made from less diluted pastes(see Fig. 5: top). As expected these paste differences were ob-served during the petrographic analysis (see below, Fig. 6).

The provenance of this tentative group is not yet definite,but it is probably somewhere in the coastal plain of southernIsrael, since reference material from Yavneh, Ashdod and Ash-kelon is quite close to it in its composition (Fig. 5; groupPALJ, Table 3: Col. 5; Goren et al., 2007). The chemical pro-file of the PALJ group was first detected in material from Qan-tir, Egypt (Fig. 1; Mommsen et al., 1996: 173, Fig. 5;Mountjoy and Mommsen, 2001: 125, Table 1: Group JPAL)and, at this time, only a probable provenance from Palestinecould be suggested, since its composition was found to beclose to a group of samples from LMLK-jars from Iron AgeIsrael, analyzed at the archaeometry laboratory in Jerusalem(Mommsen et al., 1984, 102, Table 3). Two samples fromTell es-Safi/Gath matched the PALJ chemical profile whichnow consists of 25 samples, including a possible Phoenician-

style Iron IIA ‘fluted bowl’ dated to the Iron IIA (SF158,Maeir and Shai, in press; see Alexandre, 2002; Maeir andShai, in press; see Alexandre, 2002, for a bronze bowl ofsuch form) and a sherd of a pre-LMLK jar (Fig. 3:3; SF156;see Shai and Maeir, 2003 for this type). Five similar pre-LMLK type jars from Tell es-Safi/Gath Stratum A3 were pre-viously analyzed by ICP-MS and ICP-AES (Ben-Shlomo,2006: 184, Samples SF50-SF54, Chemical Group 7, Table4.2); four of these had a similar profile comparable with theLMLK jars and the fifth was more similar to PALJ profile.A third sample, a female plaque figurine indicating a ‘Syrianstyle’ (Fig. 3:4; SF152; see Maeir, 2003b), is chemically close,but it deviates mainly in a higher cesium and a lower hafniumvalue (Table 3: Col 6) and was therefore not assigned to thisgroup. Moreover, group PALJ is still not very well defined,and it has large spreads of these two elements cesium andhafnium. Again, very large ‘dilution factor’ differences are en-countered between the samples, pointing to paste differences.If the group is divided according to these factors into three

Fig. 5. Top: Graphical comparison of chemical compositions of the group

NegA and the sherd SF160 (plotted are the differences of the concentration

values NegA e SF160 normalized by the average standard deviations. The

values of sherd SF 160 are multiplied first by the best relative fit factor with

respect to group NegA of 1.09. The concentrations are statistically similar.

Bottom: Same as top for group EME-A and sherd SF 104 multiplied first by

the best relative fit factor with respect to group EME-A of 0.89.

962 D. Ben-Shlomo et al. / Journal of Archaeological Science 35 (2008) 956e964

parts (low, medium and high concentrations), the two samplesfrom Tell es-Safi/Gath (SF156 [Fig. 3:3] and SF 158) belongto the group with high concentrations. The paste of this subgroupmatched best four reference samples of kiln wasters fromYavneh, whereas several wasters from Ashdod and Ashkelon,respectively, are made of pastes with higher amounts of admix-tures (unpublished data, in collaboration with Y. Goren). It isinteresting, that the concentrations of the LMLK group, as-signed to a probable provenance from southern Israel, possiblythe vicinity of the Lachish region (Mommsen et al., 1984: 110e2, Table 8), are still by about 10% higher. A study of larger sam-ple numbers of reference pieces from the coastal southern plainof Palestine should help resolve whether a provenance determi-nation by paste admixtures will be successful as hinted here.

Another Iron Age IIA sample (SF32) belonged to a largepithos of the ‘Ajrud’ type common in southern Israel (Ayalon,1995: 157, Fig. 8). It has a composition not very differentfrom a group ‘Edom’ (Table 3: Col. 7; Gunneweg et al.,1991: Table 4: Col. 5), but with higher potassium and rubidiumvalues. This group contains mixed Edomite pottery from theNegev and Edom, as well as locally made reference vesselsfrom Edom (Gunneweg et al., 1991: Table 2), and should prob-ably be provenanced to Edom.

A ‘White Painted’ body sherd (SF104) was assigned togroup EME-A, a chemical group believed to be Cypriot in or-igin, (Fig. 5: bottom; Table 3: Col. 8; Mommsen et al., 2006:204, Table 3: Col. 1); however, a more specific provenancewithin Cyprus is not yet possible.

The ‘Black on Red’ III juglet (Fig. 3:5; SF157, Table 2; seeSchreiber, 2003: Fig. 13:14 for this type) was not a match withany previously analyzed pottery and was the only sample fromthis study classified as unassigned (a ‘loner’). Several prove-nance studies dealing with ‘Black on Red’ pottery dated tothe 10the8th centuries have been performed (as Mattherset al., 1983; Brodie and Steel, 1996; see Schreiber, 2003:xxviexxviii). A comparison of the concentration data of thissample with the groups published by Matthers et al. (1983)was not successful. These studies have not identified the prov-enance of this ware, and it seems it represents productioncenters both in Cyprus and the Levant. While this vesselfrom Tell es-Safi/Gath was clearly not produced in southernIsrael, it is possible that future analyses may result in the iden-tification of a Phoenician, Cypriot or other provenance.

3.2. Petrographic analysis (Fig. 6)

Thin section petrographic analysis was carried out on sixsamples. The LB II Bichrome krater (SF142, Fig. 3:2; 6)was found to have been made of a brown alluvial soil, charac-terized by an optically inactive porous matrix and includesabout 25% poorly sorted angular and sub-angular silty quartz(frequency estimates of the inclusions are according to theirpercentage from slide area); it was probably fired to 850 de-grees or more. This fabric is also local to Tell es-Safi/Gathand is typical of Canaanite and Philistine vessels of the LBII and Iron Age I (Ben-Shlomo, 2006: 167e168, Group A).The plaque figurine (Fig. 3:4; SF152) was found to be madeof a fabric with a silty, rather compact matrix with inclusionsof 10% moderately sorted silty angular quartz and 5% fine-medium sandy rounded limestone fragments. Several coarsecalcareous rock fragments and rounded foraminifera are alsoevident. This fabric is quite similar to the loess type fabric(‘Shephelah Loess’, Master, 2003: 54, Fig. 4; see alsoEngstrom, 2004: 71e73) common in LB II and Iron Age I pot-tery wares at Tell es-Safi/Gath (Ben-Shlomo, 2006: 171e174,184), yet it contains unusually large calcareous rock frag-ments. While this figurine was probably made from a localclay the inclusions are coarser than other LB II samples, in-cluding a typical Canaanite plaque figurine. The red slippedincised sherd (SF149) was made of a similar calcareous loessclay fabric common to the Shephelah area, but it was finer ingrain. Its fabric includes decomposed calcite, indicating a rela-tively high temperature firing (above 750 degrees). Interest-ingly, samples SF142 (Fig. 3:2, 6) and SF149 were assignedto the same NegA group by chemical analysis, yet SF152(Fig. 3:4, 6), which resembles SF149 according to petrography(both are identified as loess soil), was found to be associated toa different chemical group, PALJ. This should not be verysurprising as these two chemical groups (PALJ and NegA)are not yet very well defined on the one hand, and the pedolog-ical attributes of the brown alluvial soil and the Shephelah lo-ess soil are similar to a large degree (Wieder and Gvirtzman,1999: 236).

Two of the Iron Age imported sherds were analyzed bypetrography; generally the matrix was too fine and the

Fig. 6. Photographs of thin sections of vessels from Tell es-Safi/Gath: SF142 (field width 6.8 mm); SF149 (field width 6.8 mm); SF152 (field width 6.8 mm); SF105

(field width 1.7 mm). All photographs are in crossed nicols.

963D. Ben-Shlomo et al. / Journal of Archaeological Science 35 (2008) 956e964

inclusions too few to draw any definite conclusions concern-ing these vessels. The White Painted sherd (SF104) hasa slightly active silty matrix with some poorly sorted siltyquartz and coarse silt-fine sand limestone fragments. TheProto-Geometric skyphos (SF105; Fig. 6, note the highermagnification) is made of a very fine well-levigated andwell-fired matrix with hardly any inclusions observed. Ina larger magnification abundant fine silty mica inclusionscan be seen; possibly this fabric can be compared to fabricsfrom the Aegean region (Whitbread, 1995: 302e313, Pl.5.23e5.25, especially Pl. 5.24, 304, ‘fine ware fabric’). Sam-ple SF32 of a large pithos resembles the Motza clay forma-tion on account of silt sized rhombs (indicating dolomiticfeatures) embedded in the matrix. Vessels from this typefound at other sites were also assigned to this fabric (Ayalon,1995: 157); this petrographic group has a very wide distribu-tion in the central hills as well as in Transjordan (see, e.g.,Goren, 1995: 291e292).

4. Discussion and conclusions

According to provenance studies of Mycenaean IIIB im-ports in the Levant, it seems that the workshops of the north-eastern Peloponnese, possibly in the region of Mycenae/Berbati, had an almost complete monopoly for the export ofMycenaean pottery to the Levant (Badre et al., 2005: 15).

This picture is not altered according to the few samplesfrom Tell es-Safi/Gath, although one of the three samples(SF163) could not be assigned and may belong to a newAegean workshop exporting Mycenaean vessels, which isnot yet known to us.

The probable local provenance of the sherd with theEgyptian Hieratic inscription has important implications(Fig. 3:1). The inscription is probably complete, reading oneword- sps(j) (‘noble, august, rich’), which indicates a dedica-tion of a sps-jar; the signs were made before firing and theirstyle dates the inscription to the 20th Dynasty (Maeir et al.,2004: 132e134). If this vessel, which was found with otherobjects imported from Egypt, was produced in Canaan, possi-bly at Tell es-Safi/Gath, it indicates a certain direct Egyptianpresence (or at least, influence) at Tell es-Safi/Gath duringthe 13th century BCE. It was possibly related to a certainEgyptian affiliated cult practices.

Two artifacts which typologically appeared to be non-local,the incised and red slipped sherd (SF149) and the plaquefigurine (Fig. 3:4; SF152), were found to be locally made inPhilistia, according to both NAA and petrography. Anothervessel, dated to the Iron IIA, which morphologically resemblesPhoenician fluted metal bowls (SF158, Maeir and Shai, inpress), was found to be locally made in southern Israel.

The Proto-geometric/Sub-Mycenaean skyphos (SF105),which is one of the few examples of this ware from the

964 D. Ben-Shlomo et al. / Journal of Archaeological Science 35 (2008) 956e964

southern Levant (and possibly the earliest import of its kind inthe region), and probably the first one to be chemically ana-lyzed, was clearly imported from the northeastern Peloponn-ese. While the sherd could be classified stylistically as eitherProto-geometric or Sub-Mycenaean, it definitely cannot nowbe defined as ‘East Greek’. The appearance of this sherd inStratum A4, dated to the late 10th or early 9th century BCEtestifies to trade connections between the Greek mainlandand the southern Levant, even in a non-coastal site as Telles-Safi/Gath (see also Waldbaum, 1994). The use of this paste,typical of Mycenaean IIIA-C vessels during the Sub-Myce-naean/Proto-Geometric period, indicates a certain continuityin pottery production techniques well into the post-palatialperiod. The White Painted sherd (SF104) is provenanced toCyprus and adds up to additional information of importsfrom this region during the Iron II. Concerning the ‘Blackon Red’ juglet (Fig. 3:5; SF157) it is not yet clear whetherthis vessel was imported from Phoenicia or from Cyprus,and only further archaeometric research of similar materialfrom Lebanon and Syria could resolve this issue.

Acknowledgment

We wish to thank the staff of the research reactor at Gees-thacht (GKSS) for irradiating the samples. In addition, wewish to thank I. Shai, A. Zukerman, R. Avisar, S. Gur-Ariehand A. Dagan (of the Tell es-Safi/Gath Archaeological Project)for assistance in the preparation of the article; and to J. Rosen-berg for drawing the map in Fig. 1.

References

Alexandre, Y., 2002. A fluted bronze bowl with a Canaanite-Early Phoenician

inscription from Kefer Veradim. In: Gal, Z. (Ed.), Eretz Zafon: Studies in

Galilean Archaeology. Israel Antiquities Authority, Jerusalem, pp. 65e74.

Ayalon, E., 1995. The Iron Age II pottery assemblage from Horvat Teiman

(Kuntillet ‘Ajrud). Tel Aviv 22, 141e205.

Badre, L., 1998. Late Bronze and Iron Age Imported pottery from the Archae-

ological Excavations of Urban Beirut. In: Karageorghis, V. and Stampolidis,

N. (Eds.), Eastern Mediterranean: Cyprus-Dodecanese-Crete. 16th - 6th

cent. B.C. Athens, University of Crete and Levantis Foundation, pp. 74–84.

Badre, L., Boileau, M.-C., Jung, R., Mommsen, H., 2005. The provenance of

Aegean- and Syrian-type pottery found at Tell Kazel (Syria). Agypten und

Levante 15, 15e47.

Beier, Th., Mommsen, H., 1994. Modified Mahalanobis filters for grouping

pottery by chemical composition. Archaeometry 36, 287e306.

Ben-Shlomo, D., 2006. Decorated Philistine Pottery: An Archaeological and

Archaeometric Study. In: B.A.R. International Series No. 1541. Archaeo-

press, Oxford.

Brodie, N.J., Steel, L., 1996. Cypriot black-on-red ware: towards a character-

ization. Archaeometry 38/2, 263e278.

Engstrom, C.M.A., 2004. The Neo-Assyrians at Tell el-Hesi: a petrographic

study of imitation Assyrian Palace Ware. Bulletin of the American Schools

of Oriental Research 333, 69e81.

Fantalkin, A., 2001. Low chronology and Greek protogeometric and geometric

pottery in the Southern Levant. Levant 33, 117e125.

Goren, Y., 1995. Shrines and ceramics in chalcolithic Israel: the view through

the petrographic microscope. Archaeometry 37/2, 287e305.

Goren, Y., Na’aman, N., Mommsen, H., Finkelstein, I., 2007. Provenance

Study and reevaluation of the cuneiform documents from the Egyptian res-

idency at Tel Aphek, �Agypten und Levante XVI, 161e171.

Gunneweg, J., Beier, T., Diehl, U., Lambrecht, D., Mommsen, H., 1991. ‘Edom-

ite’, ‘Negbite’ and ‘Midianite’ pottery from the Negev desert and Jordan: in-

strumental neutron activation analysis results. Archaeometry 33, 239e253.

Maeir, A.M., 2003a. Notes and news: Tell es-Safi/Gath. Israel Exploration

Journal 57/2, 237e246.

Maeir, A., 2003b. A Late Bronze Age, Syrian-style figurine from Tell es-Safi/

Gath. In: Deutsch, R. (Ed.), Shlomo; Studies in Epigraphy, Iconography,

History and Archaeology in Honor of Shlomo Moussaieff, Tel Aviv-Jaffa.

Archaeological Center, pp. 197e206.

Maeir, A.M., Shai, I., in press. An Iron Age IIA Phoenician-style fluted

ceramic bowl from Tell es-Safi/Gath: a ceramic imitation of a metal pro-

totype, Journal of the Serbian Archaeological Society.

Maeir, A.M., Martin, M., Wimmer, S.J., 2004. An incised hieratic inscription

from Tell es-Safi, Israel. Agypten und Levante 14, 125e134.

Master, D.M., 2003. Trade and politics: Ashkelon’s balancing act in the sev-

enth century B.C.E. Bulletin of the American Schools of Oriental Research

330, 47e64.

Matthers, J., Liddy, D.J., Newton, G.W.A., Robinson, V.J., Al-Tawel, H., 1983.

Black on red ware in the Levant: a neutron activation analysis. Journal of

Archaeological Science 10, 369e382.

Mommsen, H., Perlman, I., Yellin, J., 1984. The provenience of the LMLK

jars. Israel Exploration Journal 34/2, 89e113.

Mommsen, H., Lewandowski, E., Weber, J., Podzuweit, Ch, 1988. Neutron ac-

tivation analysis of Mycenaean pottery from the Argolid: the search for

reference groups. In: Farquhar, R.M., Hancock, R.G.V., Pavlish, L.A.

(Eds.), Proceedings of the International Symposium on Archaeometry.

University of Toronto, Toronto, pp. 165e171.

Mommsen, H., Kreuser, A., Lewandowski, E., Weber, J., 1991. Provenancing

of pottery: a status report on neutron activation analysis and classification.

In: Hughes, M., Cowell, M., Hook, D. (Eds.), Neutron Activation and

Plasma Emission Spectrometric Analysis in Archaeology. British Museum

Occ. Paper 82. British Museum, London, pp. 57e65.

Mommsen, H., Beier, Th., Hein, A., Podzuweit, Ch, Pusch, E.B.,

Eggebrecht, A., 1996. Neutron activation analysis of Mycenaean sherds

from the town of Ramesses II near Qantir and Greek-Egyptian trade rela-

tions. In: Demirci, Sx., Ozer, A.M., Summers, G.D. (Eds.), The Proceedings

of the 29th International Symposium on Archaeometry Ankara. Tubitak,

Ankara, pp. 169e178. 9e14 May 1994.

Mommsen, H., Beier, T., Hein, A., 2002. A complete chemical grouping of the

Berkeley neutron activation analysis data on Mycenaean pottery. Journal of

Archaeological Science 29, 613e637.

Mommsen, H., Schwedt, A., Attula, R., 2006. Chemische Klassifizierung von

137 Proben aus den Grabungen in Emecik und des Topfereistandortes Re-

sadiye durch Neutronenaktivierungsanalyse. In: Berges, D. (Ed.), Knidos,

Beitrage zur Geschichte der archaischen Stadt. Mainz, pp. 199e204.

Mountjoy, P., Mommsen, H., 2001. Mycenaean pottery from Qantir-Piramesse,

Egypt. Annual of the British School at Athens 96, 123e155.

Perlman, I., Asaro, F., 1969. Pottery analysis by neutron activation. Archaeo-

metry 11, 21e52.

Schreiber, N., 2003. The Cypro-Phoenician Pottery of the Iron Age. Brill,

Leiden.

Shai, I., Maeir, A.M., 2003. Pre-LMLK jars: a new class of Iron Age IIA stor-

age jars. Tel Aviv 30/1, 108e123.

Sharon, I., Gilboa, A., Jull, A., Boaretta, E., 2007. Report on the first stage of

the Iron Age dating project in Israel: supporting a low chronology. Radio-

carbon 49 (1), 1e46.

Waldbaum, J.C., 1994. Early Greek contacts with the Southern Levant, ca.

1000e600 B.C.: the eastern perspective. Bulletin of the American Schools

of Oriental Research 293, 53e66.

Whitbread, I.K., 1995. Greek Transport Amphorae. A Petrological and Ar-

chaeological Study. In: Fitch Laboratory Occasional Paper 4. The British

School, Athens. Exeter.

Wieder, M., Gvirtzman, G., 1999. Micromorphological indications on the

nature of the Late Quaternary Paleosols in the Southern Coastal Plain of

Israel. Catena 35, 219e237.

Wimmer, S., Maeir, A.M., 2007. The Prince of Safit: a Late Bronze Age hier-

atic inscription from Tell es-Safi/Gath. Zeitschrift des Deutschen Palastina-

Vereins 123 (1), 1e12.