New excavations at the Barbar Temple, Bahrain
Introduction
In the years 1954–62 a mound lying south of the
village of Barbar on the north coast of Bahrain was
excavated by the Danish archaeological expedition.
It proved to contain three temples, following each
other in time, the one built upon and around the
other, and all dating to the centuries around 2000
BC, when Bahrain was the centre of the Dilmun
culture. Substantial parts of the first and the second
temple were preserved and could be divided into
four sub-phases – Temples Ia-b and IIa-b – whereas
less is known about the third temple (III).
The architecture of the Barbar Temple consists of
three main structural parts, most clearly seen in
Temple II: 1) a high central platform with a double
altar where sacrifices took place, surrounded by an
oval platform, connected by 2) a processional stair-
case to a pool towards the west, a deep-lying
chamber with fresh water, interpreted as the dwell-
ing of the god (abzu), and by 3) a ramp leading
eastwards to a court with abundant evidence of fire,
where the sacrificed animals were cooked.
A related structure was identified and partly
excavated to the northeast of the Barbar Temple
under the designation of the Northeast Temple.
The excavations in the 1950s and 1960s at the
Barbar Temple and the Northeast Temple have
recently been published (1), and a number of
questions that emerged from the preparation of this
publication prompted a new excavation campaign
that took place from January 30th–March 31st 2004
(2). The same horizontal and vertical coordinate
systems were used in 2004 as in the old excavations.
The water channels by the pool (693)
In the south, west and north walls of the pool
chamber were four openings for subterranean chan-
nels (Fig. 1), apparently water conduits, but it was
unclear whether these channels brought water into
the pool or away from it (3). In order to clarify this
problem the continuation of the southeastern chan-
nel was investigated in two soundings during the
2004 campaign.
The first sounding was located c. 12 m southeast
of the pool (Fig. 2.1) and here the channel was
uncovered only 0.7 m from the ashlar oval terrace
wall of Temple IIb. It is clear from the section (Fig. 3)
that the channel was constructed in a large pit dug at
a time when the area in front of the ashlar wall had
accumulated to a surface level of c. 4.20. The base of
the structure was set on subsoil sand at level c. 6.25
and consisted of two rows of well-dressed ashlars
(Fig. 4), on top of which the walls were built of
In 2004 excavations at the Barbar Temple dating to c. 2000 BC established that
the function of the southeastern subterranean channel connected to the
temple pool was to supply the pool with water and not to divert water away
from the pool. North and west of the temple the oval terrace wall of its third
phase was located. In the centre of the Northeast Temple the remains of a
shaft-stairway leading down to a well-chamber was uncovered and shown to
be furnished with two subterranean channels leading water into the
chamber. South of the Northeast Temple a house with late pottery was
excavated. Two stamp seals and two seal-impressed tokens were found.
Keywords:
Bahrain, Dilmun, Barbar Temple, abzu, water channel, oval
platform
Flemming Højlund, Pernille
Bangsgaard, Jesper Hansen,
Niels Haue, Poul Kjærum
and Dorthe Danner Lund
Moesgaard Museum,
Denmark
Flemming Højlund
Moesgaard Museum
DK-8270 Højbjerg
Denmark
E-mail: farkfh@hum.au.dk
Arab. arch. epig. 2005: 16: 105–128 (2005)
Printed in Singapore. All rights reserved
105
uncut stones, 20–30 cm across, laid in clay, the roof
being made of particularly large undressed slabs,
c. 65 · 40 cm. The inner height of the channel was
c. 1.8 m, and the distance between the walls at the
bottom was c. 50 cm. At mid-height the walls were
only 15 cm apart, but this seems to be a result of
subsidence. The floor of the channel, the interface
between layers 27 and 8, consisted of subsoil sand
with an uneven surface, perhaps hollowed out
down to level 6.42 towards the north by the force
of the running water. Above that the channel was
almost completely filled with horizontal or slightly
dished layers of sand, with some pottery of Temple
IIb variety (4) (Figs 5–6). The pottery found in
the layers surrounding the channel also dated to
Temple IIb.
The second sounding was located c. 18 m southeast
of the pool, at a distance of 2.7 m from the Temple
IIb oval terrace wall (Fig. 2.2). Again the channel
was built on subsoil sand, with two rows
of well-dressed ashlars in a bottom course, on top
of which the walls were built of uncut stones,
20–40 cm across, laid in clay, the roof being in this
case made of well-dressed ashlars (Fig. 7). The inner
height was only about 1 m and the distance
between the walls 30–40 cm. The floor of the channel
consisted of subsoil sand at level 6.26, slightly
hollowed, perhaps by the running water (Fig. 8).
Above that, the channel was almost completely
filled with horizontal or slightly dished layers of
sand. No pottery was found inside the channel, but
the layers surrounding the channel contained Temple
IIb pottery.
From the opening into the pool, the first 4–5 m
of the channel can be seen to be constructed entirely
of well-dressed ashlars, built in bond with the
walls of the pool chamber (Fig. 9). In 1959 the
exterior of the channel was exposed from 2.5 to 5 m
from the pool chamber and was here also construc-
ted of well-dressed ashlars (5). The floor of the
channel consisted of subsoil sand, slightly hollowed
out, with a level of 6.61 for the first half metre.
Between the channel and the pool was an ashlar
‘threshold’ with a surface at level 6.16, and directly
above this threshold were two square holes cut in
the walls, perhaps for securing some kind of sluice-
gate.
Dating
The walls of the channel are bonded into the walls of
the pool chamber so that they must be the result of
one and the same building phase. The pool chamber
and therefore also the channel form an integral part
of the latest phase of Temple II, i.e. IIb, which was
covered up before the construction of Temple III.
The channel was built in a pit cut into deposits
accumulated in front of the oval terrace wall of
Temple IIb and must therefore belong to an
advanced phase of Temple IIb. The pottery found
inside the channel and in the layers surrounding the
channel can typologically also be dated to the time of
Temple IIb.
Conclusions and discussion
The levels of the floor of the channel measured in the
two soundings and at the opening into the pool
(from SE to NW: levels 6.26, 6.42 and 6.61) make it
clear that the floor slopes down towards the pool, i.e.
it must have brought water into it. While there is no
reason to doubt this conclusion, it nevertheless
raises some questions.
If the channel was intended to fill the pool with
water, why was the floor of the channel then level
with the floor of the pool? It would surely have
been more practical to have the channel outlet into
Fig. 1.
Barbar Temple pool with wall opening into the southeast
channel, from the northwest (1960).
F. HØJLUND ET AL.
106