THE MIDDLE BRONZE AGE
Middle Bronze occupation was discovered at Tall Safut in the 1982 and 2001 excavations
seasons, bookends for the project. In 1982, Area D was opened to investigate the earlier reports
of a glacis at the site. This area is where the majority of MB sherds were found. A few other
sherds were excavated in fill layers in areas B and C.
Prior Research
Professor Wimmer reported on the discovery of Middle Bronze sherds in his preliminary
report on the 1982-1985 seasons (1987). In his discussion of the glacis, he mentioned that only
MB/LB pottery was in the layers immediately above the glacis (Wimmer 1987: 165). He
surmised that Middle Bronze remains might underlie much of the later Late Bronze and Iron Age
remains. Wimmer would later attempt to find these remains in 2001 by digging below the late
Iron Age buildings in B4, C3, and C7 (no report from this season was ever published). It is in
these squares that the other MB sherds were discovered. Additionally, Father Roland De Vaux
(1938: 418) reported sherds from the site dating to the “Bronze I and Bronze II” periods
(although he would later rethink this position after discussing it with Glueck); and Sauer, after
sherding the site, reported that the slopes “were dripping with” Middle Bronze and Late Bronze
sherds (personal communication Wimmer; see Sauer 1986: 6).
Area D and the Glacis
In 1953 a new road from Jerash to Naur was constructed and part of the northern slope of
Safut was destroyed. In the section created by the bulldozers, Ma’ayeh (1960) reported that the
bulldozers revealed “a sloping plastered ‘glacis’ revetment, resting on natural rock, and crowned
by a wall. The glacis was constructed of different layers of sand, huwar, and earth, beaten into a
kind of terre pisee.” This report is a textbook description of a typical MBIIC rampart (Burke
2008: 51 notes that the term “terre pisee” was often misused in the literature describing rampart
construction). In subsequent years this information was cited as evidence for Middle Bronze
occupation at Safut (Ottoson 1969; Zayadin 1973; Weippert 1979; Dornemann 1983; Sauer
1986; and Mazar 1990). In 1982 Wimmer laid squares D1 and D2 in the middle of what
remained of the glacis on the south side of the tell. These squares were placed in a saddle
slightly below the remains of a wall emerging out of the tell and curving towards the east.
Unfortunately the squares were not placed directly against the wall (to better understand the
relationship between the glacis and the fortifications) for fear of destabilizing that section of
stone.
Five loci were excavated in D1, including D1.5, which was bedrock. Locus D1.1 was
topsoil and revealed Middle Bronze, Late Bronze, and Iron Age II sherds. Locus D1.2 was the
glacis itself as a 1m wide test trench was excavated along the west balk in order to reach the
glacis and no loci were designated for the dirt removed. D1.3 was located in the northern part of
the square and consisted of compact soil and limestone wall stones, collapsed from the perimeter
wall directly north of the square. D1.4 was made up of loose gray soil and was located
throughout the square directly above the glacis. It is in this locus that large amounts of Middle
Bronze and Late Bronze pottery were found. D2 was located directly to the east of D1. In D2
five loci were excavated, the contents of which corresponded nicely with what was found in D1:
D2.1 corresponds with D1.1 in that it contained topsoil; D2.2 corresponds with D1.4 in that it
contained lose gray soil and was located above the glacis (it is this locus from which the MB
sherds located at Andrews University come); D2.3 corresponds with D1.2 which is the glacis;
D2.4 corresponds with D1.3 which contained the perimeter wall collapse; and D2.5 equals D1.5
which was bedrock.
The loci designated as the “glacis” are D1.2 (Munsell Reading 2.5Y 7/4 pale yellow –
10YR 8.2 white) and D2.3. In D1 and D2, this glacis was exposed over both squares, except in
the southwestern corner of D2 where bedrock was exposed below the gray soil and along the
northern balk of both squares where wall collapse had accumulated. The slope of the glacis was
listed at 45 degrees, which is well above the average of a manmade glacis (Burke 2008: 50,
Table 5). The square supervisors and area supervisor all described the makeup of the glacis
differently, making it difficult to determine the actual material. In D1.2 it was described as “[a]
very compact clay layer,” “a limestone cap” and “a natural structure” (Wimmer, R. and Gajary
1982: 3-4, 23). The square supervisors also noted that this locus was naturally slippery, was
somewhat “stepped” and was embedded with large rocks. In D2.3 the glacis was described as “a
limestone matrix deposit. It is compacted clay (huwwar) and pebbles (approx. pea sized)”
(Boutsikaris and Wilson 1982: 7). The supervisors also mentioned that the glacis appears to be
natural because there is no evidence of occupation above or beneath it (i.e. flat lying sherds), and
that under the wall tumble (D2.4) there was soft, fine soil and the bedrock, a fact which seems to
confirm that the glacis is natural.
It should also be noted that the area D supervisor Jennifer Groot understood this “glacis”
to be artificial, a plaster coat to keep the bedrock from eroding (Groot 1983: 2), (although at
times she called the glacis a “compacted gravel and huwwar surface” and “sterile huwwar”).
Groot dated the glacis to the Middle Bronze Age based on the similarities between this “plaster”
layer and similar features of the glacis found at Jericho, Taanach, and Tell Dan (Groot 1983: 3,
cf Pennells 1983: 58). However, these three glacis are all made out of different material: the
Jericho glacis out of mudbrick and huwwar plaster (Kenyon 1981; Marchetti 1998), the Ta’anach
glacis out of clay and huwwar limestone (Lapp 1964; 1967), and the Tell Dan galcis out of
crushed travertine (Biran 1994: 59-63). Also, the glacis at all three sites were constructed on
earthen ramparts, making the parallels to Safut even less accurate. Burke (2008: 11) notes that
the presence of a glacis is not sufficient to date fortifications to the MBIIC.
From the excavations carried out in area D, it is clear that the “glacis revetment” that
Ma’ayeh wrote of did not exist. He mistook the natural bedrock stratigraphy along with the
layers of deposition and crowning wall for a traditional Middle Bronze fortification. However,
the excavations did reveal some kind of glacis, considered natural by Wimmer and the square
supervisors, but artificial by the area supervisor. Several factors indicate that this formation of
huwwar/compact clay/plaster is most likely natural. The strongest clues are the rocks that were
protruding out of the layer, the stepped nature of the glacis, and the seeming disconnect between
the glacis and the fortification wall (since the wall collapse in both squares was on top of fine
soil and covering bedrock). As a whole, these factors seem to indicate that this glacis is in fact a
natural accumulation of friable limestone above the bedrock. However, because of the
imprecise designation for the glacis material and the lack of excavation along the fortification
wall, it is difficult to say with certainty if the glacis was natural or artificial and what its function
might have been.
Nelson Glueck (1939: 191) observed that the western and southern slopes of Safut were
more gradual in descent, and it is these two sides of the tell that would require a glacis for added
protection. Unfortunately, the construction and expansion of the Amman-Jerash highway has
eliminated the possibility for further research on the south side, but it would still be possible to
excavate on the west side of the tell. The question of the Safut glacis will not be completely
answered until this excavation can be carried out.
Areas B and C
In the 1982 season, Middle Bronze remains were found in one square outside of Area D,
B2. This square is located north of area A outside the city wall, which Wimmer dates to the
Late Bronze Age (1987: 165). One sherd from B2 locus 25 is in the collection located at Andrews
University, but records for this locus do not exist. Squares 2 and 5 of Area B are separated by
this Late Bronze wall, and it appears that this sherd came from excavation of clearing soil from
on top of the wall. There is a locus sheet labeled “Area B Square 2-5”, where the “balk”
between the two squares on top of the wall was excavated until mudbrick was found. The other
possibility is that loci 24 and 25 were excavated but locus sheets were either never filled out or
are now lost. Other Middle Bronze sherds were reported by Wimmer in the loci contiguous to
the “foundation trench” outside the city wall in B2 (1987: 165). Several sherds were drawn, but
they are forms that could date either to the Middle Bronze Age or to the Late Bronze Age.
All other Middle Bronze sherds housed at Andrews University are from the 2001
excavation season. Three loci in B4 contained Middle Bronze sherds (303, 313, and 314). Locus
303 consisted of what seems like interseasonal debris, also associated with locus 301 (which is
located above 303) from wall collapse in B4 and B7. B4 had been previously excavated in 1983
and, according to supervisors’ notes, locus 303 represented the level previously reached (B4
notebook 2001: 136). Locus 303 was located in the northeast quarter of the square. Locus 313
was an arbitrary change from Locus 312, consisted of fill, and was located in the northwest
corner of the square. Locus 314 was located below 313 and consisted of similar material. These
three loci also contained much pottery dating to the Late Bronze Age, Iron Age I, and some
dating to the late Iron Age II.
For Area C, Middle Bronze sherds were located in loci 402 and 410. Locus 402 was the
first to be excavated after cleanup and was located in the northwest corner of the square
abutting wall 403 to the west, wall 404 to the east, and wall 411 to the north. It consisted of
tumble, as other sherds found within the locus dated to the Iron Age I-II. Locus 410 also
consisted of tumble and was also located directly under the cleanup locus.
In square C7, Loci 407 and 412 revealed one MB sherd each. Locus 407 consisted of
mudbrick detritus, located under a small wall (405) and a burnt area (406). Most other sherds
from this locus were Late Bronze and Iron I. Locus 412 was soil from the creation of a sub-balk
between C3 and C7. Most of the remains appear charred and pottery dated mainly to the Iron
I. Unfortunately, the sheets for locus 415 are mostly blank. This locus is in some way
contiguous with loci 414 and 416, and most sherds, again, date to the Late Bronze and Iron Age.
There is no sheet for locus 422.
Pottery Discussion
There are twenty-four Middle Bronze sherds from Safut housed at Andrews University.
Many more were documented in pottery readings and some were reported as having been
drawn, but no other drawings could be definitively dated to this time period. Ten of these
Middle Bronze sherds were recovered from the 1982 season and fourteen from 2001. From
D2.2, there are nine sherds: five platter bowls, three carinated bowls, and one jug. There is one
jug sherd from B2.25. The number of Middle Bronze sherds from 2001 is more complete than
the number from 1982, because all diagnostic sherds from that season are currently at
Andrews. Four sherds were found in B4: one platter bowl was found in locus 303, one platter
bowl similar to that found in locus 303 and one jug (or possibly bowl) were found in locus 313,
and one small handle was found in 314. In C3 loci 402 and 410, were found one jug similar to
MBIIC types with everted triangular rims and one juglet with a white slip and burnish. In C7,
eight sherds were found. One sherd was unearthed loci 407 and 412: a storage jar and a juglet,
respectively. In C7.415, four sherds were found, a cooking pot, a jug, and two juglets. In locus
422, two sherds were found: a cooking pot and platter bowl.
Overall, the MB sherds represent eight platter bowls, three carinated bowls, five juglets,
five jugs, one cooking pot, and one storage jar. Even though the amount of pottery is small,
there is a wide representation of forms. The eight platter bowls fit into three general rim types.
Four were simple, slightly upturned rims. The other two are very similar, the first can be
described as slightly inverted and triangular (SFT01.B4.313.32.3), and the second as thickened
and rounded (SFT01.B4.303.30.1). The three carinated bowls all have open, diagonally everted
rims. Only one of the juglet sherds had a rim (SFT01.C3.410.22), its form being upright and
slightly turned in. The two jug rims are each a different type, with one being everted, triangular
(SFT01.C3.402.5, but with a unique groove) and the other flaring, simple (SFT01.C7.415.57.5).
The two cooking pot rims were everted, triangular (SFT01.C7.415.57.2) and out-turned, simple
(SFT01.C7.422.60). Eleven of the sherds (46%) can be classified as some version of Chocolateon-White ware (CW) and another six sherds have some kind of paint on them but do not fit the
classification requirements for CW (Fischer 1999). These numbers indicate that 71% of the
Middle Bronze assemblage from Safut is decorated and can be considered fine ware. However,
since sherds from the earlier seasons were only kept selectively, no definitive statements can
be made on the frequency of decorated wares in the overall assemblage.
The ceramic assemblage described here best fits in the Middle Bronze IIC. Based on
parallels, some of the sherds can date from the Middle Bronze IIB through the Late Bronze IA.
Since all sherds were taken from fill layers or layers that cannot be associated with any
architecture, it is difficult to narrow the chronological range. However, the majority of sherds
do have parallels in the MBIIC. The ceramic assemblage, especially the painted wares, is best
compared to phases IV/1-V at Tall Abu al-Kharaz (Fischer 2006), which date to the MBIIC-LBIA
(17th-mid-16th centuries BCE). There are several sherds from Safut that fit in Fischer’s
Chocolate-on-White typological groups (Fischer 1999; 2006). Eight sherds fit Fischer’s
Chocolate-on-White I and/or II (CWI/II) (Fischer 2006: 264-266). CWI/II is characterized by
monochrome decorations with a white slip and burnish, and the chequer and ladder pattern
became popular (SFT01.B4.313.31 is a good example of the chequer design). Two sherds belong
to Fischer’s Chocolate-on-White Bichrome (CWB) (Fischer 2006: 264). CWB is characterized by
chocolate-brown and black decorations and wheel burnishing: however, no bowls were found
in the Tall Abu al-Kharaz assembly, bringing the designation of these two bowl sherds as CWB
into question. There is also one sherd that could be Chocolate-on-White III (CWIII) (Fischer
2006: 266). CWIII has a much thinner slip than the earlier CWI and CWII, but it is still burnished
unlike other Late Bronze painted forms. Other sites in Jordan with many parallels to the Safut
assemblage include Tall al-Umayri (phases dating to the MBIIC, see Herr et al 2002) and Deir
Alla (Stratum IV dating to the late MBIIC, see van der Kooij 2006). Sites in Israel and Palestine
where many parallels were found included Dothan (Master et al 2008), Jericho (Kenyon and
Holland 1982; Kenyon 1986), Megiddo (Loud 1948), and Shechem (Seger 1974; Cole 1984).
Summary
What was thought to be a MB glacis was uncovered in Area D, but it appears to actually
be a natural formation. Despite the fact that there is no “glacis revetment” at Safut, there are
other remains from the Middle Bronze Age. Sherds were found in the 1982 and 2001 seasons
dating approximately to the Middle Bronze IIC. Over 70% of the sherds were Chocolate-onWhite and painted ware. Unfortunately, the loci in which they were found consisted of fill and
were not associated with any architectural features. Nevertheless, the quality of these sherds
hints at the potential occupation levels yet to be discovered.