Arabian archaeology
and epigraphy
/%,V ,'.'lT.;I'-'6
Archaeobotanical evidence for
early Dilmun diet at Sax, Bahrain
MARK NESBITT
Cambridge, UK
A flotation machine was used to process large quantities of earth at thc S a x
excavation in the 1990 and 1991 seasons. Carbonised seeds and charcoal were
rccovcrcd from A wide range of contexts dating to about 1900 BC. While ovcrall
quantities were low, enough contexts were productive to allow quantification.
[late stones were the most freq~~ent
crop remains, with smaller amounts of freethreshing wheat and hulled six-row barley. This confirms evidence from other
sourccs (textual, dental) for the importancc of dates as a staple food in the Early
Dillnun period. A survey of ethnographic and archaeological evidence for date
husbandry in Bahrain suggcsts that the date-palms and cereals were grown in
irrigated date gardens similar to those found today.
Since the 1960s there has been a remarkable increase in archaeological activity in the Arabian Gulf, but this has rarely been accompanied by systematic recovery of plant remains or bones. Lorenzo Costantini's studies on imprints and some flotation samples in
the southern Arabian peninsula, and the beginning of flotation at
Failaka in Kuwait in 1988 are the only systematic attempts at recovering plant remains so far. A large-scale programme of drysieving and machine flotation was therefore a high priority from
the first season at Saar in 1990 onwards. Very large quantities of
bone and shell have also been recovered. The amount of charred
seeds and charcoal is small but represents the only substantial data
on Dilmun plant foods and environment to be studied since the
discovery of this civilisation, some 30 years ago.
The excavations al. Saar of the London-Bahrain Archaeological
Expedition are directed by Robert Killick, Jane Moon and Harriet
Crawford. The site lies in the northwest part of the island, bounded
to the north by the houses and date gardens of Saar village, and
to the south and west by a large area of Early Dilmun burial mounds
('1 1.
EARLY DILMUN DIET ET SAAR
Recovery
The rainy winters of Bahrain are sufficient to ensure that desiccated
ancient material does not generally survive. All the plant remains
found are charred, although some pieces of charcoal appear silicified.
As no burnt destruction levels have been found at Saar, most of
the charred material came from the flotation of contexts containing
dispersed, redeposited plant remains. Without flotation, the only
plant remains to have been recovered would have been a few date
stones and fragments of charcoal from the dry-sieving; all the cereal
grains and nearly all the charcoal came from flotation.
A Siraf-type flotation machine (2) was used. As all water had to
be tanked into the site, a recycling system was used for the machine.
The outflow passed into two settling tanks, with a pump tahng
water from the second tank back into the flotation machine. This
recycling worked very well, with use of water restricted to cleaning
out and refilling the flotation drum after every 220-240 litres of
soil had been processed. The opportunity was taken to completely
clean and refill the drum and settling tanks every time the water
tanker came (c. every 5 days), but this could have been done much
less frequently if water had been in short supply. Given that all the
samples were of the same period, it was not felt necessary to empty
and clean the drum between each sample. Cross-contamination because of the water recycling also appeared minimal. Although some
tiny fragments of charred material passed through the sieves into
the recycling tanks, these formed a scum on the surface, well above
the level of the pump inlet. In any case, I doubt whether wet carbon
would survive passage through a pump.
In the 1990 season a single, 0.41 mm mesh sieve was used. In
1991 I tried using a 1 mm sieve above a 0.3 mm sieve. This is the
standard used at most Near Eastern sites, but at Saar the silt in the
deposits quickly blocked the smaller mesh. A 0.41 mm mesh was
then used instead of the 0.3 mm sieve. While this combination
worked well, examination of some of the 0.41 mm mesh fractions
from the more productive samples found no identifiable charred
remains. While examination of all these samples might have produced one or two weed seeds, it was obvious that at this site
virtually all the charred material was confined to the I mm mesh.
The decision was therefore taken to discard all the 0.41 mm meshes
where a pair of 1 mm and 0.41 mm sieves had been used and
concentrate efforts on the I mm samples.
Sampling
In 1990 the flotation machine was operated by Rebecca Montague.
Sampling was mainly restricted to deposits thought to be of ashy
or burnt appearance. I briefly examined the I990 material to see if
MARK NESBlTT
there were any significant differences to the 1991 samples, on which
analysis was concentrated. Table l shows total amounts of carbon
in the 1990 samples (measured to 0.1 g accuracy only). The main
feature of interest is two large samples of charcoal, but otherwise
the material is comparable to that recovered in 1991.
In 2993 a much wider range of deposits was sampled, but with
some variation in selection criteria between trenches. The sampling
policy used has worked successfully at other Near Eastern sites.
Essentially, at least a part of all deposits were floated, but with the
partial exception of collapse debris. In all, nearly 7,000 litres of soil
were floated in the 1990-91 season from 82 distinct contexts.
All the samples come from deposits with Early Dilmun Barbar
ware, very similar to Period 1 pottery at Failaka and thus dated to
about 1900 BC (3).
Table 2 shows the weights of the different types of charred
material, listed by unit. Where a number of samples were taken
from one feature ("unit"), scores have been amalgamated from all
the productive samples within that feature.
At Saar most of the soil excavated is in fact collapsed walls
covered in wind-blown sand. Of the 25 contexts sampled that could
be classed by the trench supervisor as collapse, or deliberate makeup (i.e. dumping of near-sterile material to level floors and suchlike), both shown as "rubble" in Table 3, only five contained any
charred material, and the quantities are (unsurprisingly) very small.
Sampling of this context type was therefore ended.
-1nhle 7. Saar 1990: charred plant remains. Data from all 1990 samples containing 0.4 g or more charred material. Total
(least
weights to an accuracy of 0.1 g only. Relative abundance of different materials ranked within samples from
abundant) to + + + (most abundant).
+
Square Unit
Object Unit description
E18
E18
F17
001
001
007
006
001
00'1
00 1
00 1
001
001
00 1
00I
003
005
005
F18
F18
F18
G1 7
G'I 7
G'I 7
G17
G17
G 18
K16
K16
K17
009
020
026
029
050
062
027
029
03 1
032
034
0'13
008
018
012
Occupation deposits
Occupakion deposits
Occupation deposits
Occupaliondeposits
Floor
Occ~tpationdepostis
Pit fill
Pit fill
Pit fill
Same pit
Pit fill
Occupation deposils
Occupation deposils
Pit fill
Pit fill
Quantity
(litres)
50
50
16
50
50
17.5
21
18
26
23
50
4 1.5
50
53.5
63.5
Dates
+
+
+
+
+
+
+-F+
+
+
+++
+
+++
++ +
+++
Charcoal
Indet.
material
+++
+++
-i-
+++
i+
+++
+++
++
+T
+++
+
-1+
-I-
+++
++
++i
+
-I-
Total
weight
3.1
1.3
0.4
1
1.8
8.7
0.5
0.8
0.5
1 l .3
'1 .S
5.2
0.4
3.2
2.6
g carbon
litre
X
1000
62
26
25
20
36
4.97
24
44
19
49 1
30
'1 25
8
60
41
'Ikhle 2. Saar 1991: charred plant remains.
Weight of charred material (g)
Square Unit
027
025
052
040
070
039
002
Date
Object taken
Unit description
23.01.91 Animal disturbance above
pit Q20:032
003
21.02.91 Waterproofed basin
001
17.03.91 Fill of plaster basin in house
001
13.02.91 Bumt area (it1 sih7) in house
001
12.03.91 bz-sitrr burnt area in small
room
002
14.03.91 In-sifu burning, othcr dcpression
023-25 29.01.91 Midden
28.02.91 Midden
05.03.91 Midden
10.02.91 Midden
20.02.91 Occupation deposits
13.02.91 Thick ashy layers W of altar
04.03.91 Temple sondage
09.03.91 10th spit in temple sondage
26.02.91 Occupation debris
02.03.91 Occupation debris in house
25.02.91 Ashy fill of pit
26.02.91 Ashy fill of pit
03.03.91 Ashy 811 of pit
03.03.91 Pit fill
12.02.91 Pit fill
04.02.91 Pit fill
05.03.91 Pit fill
13.03.91 Pit-fill
19.02.91 Pit-fill, in-sitzc burning
23.01.91 Sand collapse
24.01.91 Floor make-up
10.03.91 Floor make-up
27.02.91 Sand layer under collapse
28.02.91 Below floor (under
Q20:052)
12.03.91 Tanoor
Unit type
001
Unknown
Basin
Basin
Burnt
Bumt
Burnt
Midden
Midden
Midden
Midden
Occupation
Occupation
Occupation
Occupation
Occupation
Occupation
Pit
Pit
Pit
Pit
Pit
Pit
Pit
Pit
Pit
Rubble
Rubble
Rubble
Rubble
Rubble
Tanoor
Quantity
(litres)
Dates
Indet.
Charcoal material
Grain
Total
weight
g carbon/
litre
X 1000
MARK NESBITT
Tnble 3 Distribution of 1991 charred matenal in ddfcrcnt contexts.
Burnt
patches M~dden Pit Basin Occupation Rubble Tanoor Unknown
3
6
15
4
21
25
8
1
Number of contexts sampled
Number containing charred
3
4
9
2
6
5
1
1
mater~al
Total
83
31
Average
Presence expressed as percentage 100
Average wt (g) charred materials/
litre X 1000
36
67
60
50
29
20
13
-
37
30
23
19
10
6
36
8.5
19
Disappointing results were also obtained from generalised "occupation" contexts, i.e. the build-up of soil above floors that accompanies occupation of an area. However, some of these spreads
of soil were very rich in bone and shell, and are separately classified
as "middens" in Table 3. These areas, interpreted as refuse disposal
areas were, as would be expected, amongst the most productive in
charred seeds and charcoal.
Three spreads of debris showed signs of in-sifu burning ("burnt
patches") and all three were amongst the richest samples. In contrast
to these spreads of material, the remaining classes of deposit were
in easily defined features such as pits and basins. Nine of the 15
pits sampled were productive, as were two of the four waterproofed
basins. Surprisingly, only one of the eight tann~rs(ovens) contained
charred material.
The quantities of material are too small to allow investigation of
the spatial distribution of the plant remains, except in terms of
general categories of deposit. All the excavation trenches produced
similar types and quantities of charred plant remains.
The comparative densities from different types of contexts suggest that in future seasons, a more limited programme of sampling
should focus on areas of in-situ burning, middens, and features such
as pits and basins. Ordinary occupation deposits and tannurs are a
low priority.
Why so few seeds and charcoal?
At a typical second millennium site in the Near East, 1 would expect
a Rotation programme in which 6804 litres of soil was floated (as
in the Saar 1991 season) to produce several large boxfuls of charred
remains. The yield from the 1991 season at Saar was a mere 82 g.
Some possible explanations can be ruled out. The flotation machine
functioned very well, and there was no obvious damage to the
charred remains in the water. Inspection of the heavy residues,
including detailed sorting of the most productive samples, demon-
EARLY DILMUN DIET ET SAAR
strated that very little material was sinking. It seems then that there
are very few charred remains in the soil matrix itself.
Not only is the amount of material rather small, but its contextrelated distribution, described above, has some curious features. For
example, on a typical site in Turkey or Iraq, the pits would contain
the greatest densities of charred remains, because their final use is
often for refuse. At Saar some of the pits did appear to be full of
sand only, but others should have contained more charred refuse.
Only one of the tannurs contained any charred material - but
similar tannurs at other sites always contain some seeds and fuel
remains at the bottom.
The lack of material from Saar cannot be ascribed to limited
sampling - some 82 contexts were sampled as 123 separate samples,
nor to inadequate volumes of soil, with a minimum sample size of
40-60 litres. That even the most likely deposits were low in seeds
and charcoal suggests that there must be a general explanation, one
that applies to the site as a whole.
Little is known about the reasons for variation in seed densities
between sites, which can often be very marked. Ethnographic and
archaeological observations of the mechanisms by which seeds and
wood are preserved on sites suggest the following possible explanations:
I. Little plant material was used at the site. This seems unlikely.
O n the grounds of ubiquity in the samples that were productive,
it seems likely that dates and cereals were important foodstuffs, and
it is difficult to imagine a totally animal diet for this substantial
second millennium town. The abundance of grinding and pounding
stones, and the presence of an oven in most houses, also suggest
a typical middle-eastern, plant-food-based diet.
2. Little of the plant material present at the site was carbonised.
Even if the cereal grain was imported from elsewhere, and the dates
processed off-site, one would still expect the casual charring o f
material to occur in everyday life. At contemporary sites elsewhere
even the least productive deposits will contain a few grains - an
index of just how pervasive the charring of seeds is in an environment of hearths and ovens. There is plenty of evidence of cooking
at Saar, so the necessary fires cerkainly existed.
3. Little of the carbonised plant remains were incorporated into
the soil matrix of the site. There are two possibilities here. The
inhabitants of Saar may have taken great care to use all ashy refuse
(as opposed to bone, which they left all over the site) as fertiliser.
As Popenoe (4) points out, the desert soils of date gardens in
this region typically lack humus and benefit from ample fertiliser.
Anything available may be used, ideally manure, but also fish heads
and domestic refuse. The relative "cleanliness" of the houses, with
few artefacts or layers of "fill" might support this idea. The other
MARK NESRITT
possibility is that ashes were dumped iil open areas and were broken
down by climatic effects and trampling before becoming sufficiently
covered by soil to be protected. Perhaps the stone construction of
the site and the sandy nature of the soil prevent the build-up of
the kind of protective, washed out mud-brick type of deposit found
in countries further north?
4. Post-deposition destruction. The occupation levels of the site
are covered in collapsed debris to a thickness of up to 1 m. Carbon
preservation should be excellent under this cover, and there are no
obvious soil conditions (such as salinity) at the site that might result
in seeds and charcoal disintegrating. However, little is known of the
effects of soil conditions on the preservation of charred materials.
While I suspect the answer lies with the incorporation of charred
material into the site deposits, a definite explanation will need to
await study of the site formation processes by soil micromorphology, and their comparison to other sites. George Willcox
(5) remarks on the similarly fragmented and scarce nature of the
plant remains from Failaka, and it may be that preservation is similarly poor at most Gulf sites.
Identification criteria
Date (Phoenix ductylifera)
Whole date stones, with their longitudinal ventral groove and
round embryo spot on the dorsal side, are easy to identify (Fig.
I). Fragments show a distinctive structure in transverse section of
0
cm.
5
Fig. l .
Date stones from dry-sieved deposits, Saar 1991.
10
EARLY DILMUN DIET ET SAAR
Erblr 4. Measurements of charred dake stones from Saar and Failaka.
Measurements (mm)
L
Sample
SAAR 1990
K17:12:05
W
11.9
T
L:W
T:W
5.4
SAAR 1991
E16:06:03
E16:10:05
E16:13:04
E16:14:01
E16:16:01
EI6:17:03
Q20:32:05
R20:28:04
Saar
Min.
Avg.
Max.
n.
Failaka
Min.
Avg.
Max.
n.
10.4
13.33
19.5
16
46
5.97
7.2
15
5.0
5.66
6.3
16
'153
225
285
15
83
96
113
15
14.8
16.95
18.7
4
5.7
6.51
7.5
28
4.8
5.85
6.8
28
256
268
280
4
81
90
98
28
Failalta measurements from Rowly-Conwy, Remains of date (Pl~oet~ix
dactylifer~~)
from Failaka.
Dote stones
Width versus thickness
4-64
4.5
5.0
5.5
6 -0
6.5
Width (mm)
Saor
0 Failaka
Fixix 2.
Width and thickness of date stones from Saar and Failaka.
7.0
I
7 .S
MARK NESBITT
radiating cells. A few stones have traces of the papery skin (endocarp) of the seed, and one has charred with the flesh almost intact.
The size of the Saar date stones is very similar to those of Failaka
(Table 4; Fig. 2). The variation in length - 10.4 to 19.5 mm - is
particularly striking, and it is possible that, once larger numbers of
stones are available for measurement, clusters of different sizes
conforming to varieties may become apparent.
The hard, woody cup-shaped perianth usually adheres to the
date fruit after picking, and it is surprising that none are present in
the samples. They are too knobby to eat comfortably, and if the
date stones result from domestic consumption, one would expect
the perianths to have been discarded and charred too, as they were
at Larsa (6).
Wheat (Tviticum durumlaestivum) and barley (Hordeurn vulgare)
The seven wheat grains all have the rounded flanks and greatest
width close to the embryo typical of free-threshing wheats (Fig. 3a,
a. S~nallfree-threshing wheat grain. Saar 1991. K18:19:02. b. Large free-threshing
wheat grain. Saar 1991. L17:46:01. c. Bread wheat ('1: uestivum) rachis internode.
Saar 1991. K18:20:11. d. Twisted hulled barley grain. Saar 1991. L18:3'1:01.
EARLY DILMUN DIET ET SAAR
b). O n e wheat rachis (Fig. 3c) has the shield-shaped rachis, longitudinal striations and low-rimmed cups at the glume bases of hexaploid
(bread) wheat ( T acstivum). However, with only one rachis it is
impossible to be certain that all the grains derive from hexaploid
wheat.
The barley husks have not survived charring, but the angular
corners of the grains in transverse view demonstrate that the barley
is a hulled type (Fig. 3d). Nine grains are definitely twisted and,
nine straight. The presence of twisted grains shows that 6-row
barley is present. While the ratio of twisted to straight grains is
not precisely the 2 : 1 to be expected of pure 6-row barley, the
small number of grains and the high proportion too distorted to
be classed as twisted or straight, mean that a deviation from the
2 : 1 ratio need not be evidence for the additional presence of 2row barley. The barley rachis internode comes from the basal part
of the rachis and is not referable to type.
Importance of the plant foods
With a larger suite of plant remains, one would typically analyse
each sample in terms of its place in the crop-processing sequence,
and thence of the relative abundance of each species in each sample.
It would then be possible to characterize each species as major crop,
minor crop, contaminant, import, etc. The quantities of material
from Saar are too small to allow such detailed studies.
However, although the quantities are small, we do have a suite
of 31 samples, and 82 g of charred material (Table 5). If these seed
remains have any relationship to plant use on site (and I believe it
to be a severely "filtered" but representative set of material), then
the relative presence of species in those samples that contained
charred material, and the abundance of the species overall, can be
used (cautiously!) as an indicator of overall importance at the site.
As date stone fragments are present in 3 0 of the samples, and are
the most abundant seed in terms of volume and weight, it is likely
that dates were an important foodstuff. Similarly the presence of
barley grains in 58% of samples indicates that it was important,
while naked wheat (present in only 13%)was less important.
'lirblc 3. Relative abundance of the 1991 ~ l a n remains.
t
Date
Presence in contexts (n=31)
Presence expressed as percentage
Total quantity (g)
Indet.
Charcoal material
Any
Barley
Wheat
cereal
Weeds
30
29
6
97
94
19
18
58
4
13
20
65
23
66.11
7.17
8.65
-
-
-
-
7
MARK NESBITT
While this is a narrow range of crop plants, with no other fruit
or vegetables recovered, ethnographic data from the Gulf discussed
later on does support an interpretation in which dates, followed by
barley and wheat, are the major food items.
Uses and history of the plant foods
Cereals
The presence of six-row hulled barley and free-threshing wheat is
not surprising - both are present at sites of the same period in Iraq.
The number of grains is too small to allow us to be absolutely sure
that the absence of emmer is significant, but if it was important in
Bahrain at the time (and it is at Mesopotamian sites), one would
expect at least some grains here.
The presence of two pieces of cereal chaff does not necessarily
indicate local crop processing: some cereal chaff invariably remains
behind in processed, cleaned stored grain. While barley is most
commonly noted as animal feed (7) in cuneiform documents, both
the barley and the free-threshing wheat present at Saar are suitable
for a wide range of foodstuffs, and the presence of a tannur in
every house suggests that flat bread (such as that found at midthird millennium Ur (8)) was one. The husks of hulled barley can
easily be removed on a grinding stone so that the grain is suitable
for human use (9). The greater abundance of barley compared to
wheat reflects the same pattern in Mesopotamian texts and archaeobotanical assemblages.
History of the date
Little is known about the domestication of the date palm. Efforts
to distinguish truly wild palms from feral escapes have met with
limited success (IQ, but the original wild ancestor would have
grown somewhere in the area of North Africa, Arabia, the southern
part of the Near East and the Indus basin where dates are cultivated
today. It is perfectly possible that truly wild dates once grew on
the swampy shoreline of Bahrain, but the cultivation of date gardens
over millennia would have wiped out any trace of their existence.
Many archaeobotanical records of date stones are from earlier
excavations where the stratigraphy is unreliable. Records from early
sites that do seem reliable include:
Tepe Gaz Tavila, Iran
Tell el-Oueili, Iraq
Eridu, Iraq
Teleilat Ghassul, Palestine
Jericho, Palestine
5400-4800 BC (11)
5th mill. BC (12)
4th mill. BC (13)
3700-3500 BC (14)
3200 BC (15)
EARLY DILMUN DIET ET SAAR
Nahal Mishmar, Palestine
Hili 8, U.A.E.
Tell Abraq, Oman
Ur, Iraq
Tepe Yahya, Iran
Shahr-i Sokta, Iran
Failaka, Kuwait
Ar-Raqlah, Yemen
Tell Yelkhi, Iraq
Tell ed-Der, Iraq
Qal'at a1 Bahrain, Bahrain
Rumeilah, U.A.E.
3200-3000 BC (16)
3000 BC (17)
2600-2500 BC
late 3rd-early 1st mill. (18)
mid-third mill. BC (19)
2400-1800 BC (20)
2100 BC (21)
2000 BC (22)
2000 BC (23)
2000-1800 BC (24)
Early 2nd mill. BC (25)
1475 BC (26)
800-400 BC (27)
Costantini (28) reports two uncarbonised, silicified date stones from
Mehrgarh in Pakistan, dating t o 6000 and 5000 BC. It is always
difficult to be certain of the age of uncharred material, which may
be intrusive. Confirmatory material would be very welcome.
While the archaeobotanical record is doubtless very incomplete,
and likely to remain so until dry-sieving or flotation is more widely
implemented, it does seem that dates have been cultivated since at
least the fifth millennium BC.
Dates in Mesopotamia
In addition to the archaeobotanical evidence cited above, there is
an impressive corpus of seal engravings and texts documenting the
use of the date palm in Mesopotamia since the middle of the
Sumerian period (c. 2500 BC onwards). It should be noted that this
is simply the first date for which we have this kind of evidence in
quantity; the archaeobotanical evidence is that dates, whether wild
or cultivated, were used in Mesopotamia well before we have
written or pictorial evidence for them.
The cuneiform texts that mention dates are mainly contractual,
therefore relating either to date palm husbandry, or to the supply
of dates in baskets. The most common terms are "suluppG" and
"uhinnu". Both terms are used in Old Akkadian language (c. 2500
BC) onwards, and their precise meanings are still unclear. The Chicago Assyrian Dictionary (29) suggests "ripened and plucked dates
for "suluppQ", Postgate (30) suggests "dried dates"; for "uhinnu"
von Soden (31) has "fresh, green dates", Postgate "fresh date, in
the autumn".
There are few clues to culinary uses. Landsberger (32) quotes a
recipe for a cake "but6" made from dates, pomegranates, raisins and
figs, and the Chicago Assyrian Dictionary (33) defines the NeoBabylonian (c. 1000 BC) term "gidd@" as a kind of cake made from
MARK NESBITT
dates, emmer and sesame. There is also a brief textual reference to
a wooden mortar used for processing dates (34), tantalising in view
of the abundant evidence for both dates and milling equipment at
Saar. In second millennium Mesopotamia, barley beer was the main
alcoholic drink, and date wine does not become important until
the mid-first millennium BC (35). The textual evidence is therefore
unhelpful, beyond suggesting that the "seated god in a flounced
robe drinking through a straw" found on one of the seals from Saar
(36),is more likely to be drinking a cereal beverage than date wine.
Cuneiform evidence for Dilmun dates
The term "asnf" is used in Old Babylonian (c. 2000 BC) documents
to refer to "Dilmun" date palms and dates (37). Uses mentioned in
the texts include temple offerings and food. As Potts (38) points
out, "asnQ" refers to a variety of date palm rather than to dates
imported from Dilmun. "AsnQ" date palms appear to have produced
especially sweet dates. That the reputation of "Dilmun dates" reflects the renown of dates from Dilmun itself is suggested in a
Sumerian myth quoted by Kramer (39):
"Dilmun - its dwellings are good dwellings,
Its barley is very small barley (40),
Its dates are very large dates."
Uses of the date palm in Bahrain
Given that dates were a major resource in Dilmun period Bahrain,
how might they have been used, and how would this be reflected
in the archaeological record? Although dates are no longer a major
foodstuff in terms of calories, and the dried dates on sale in the
Manama market today come from Saudi Arabia, dates are still
highly regarded by Bahrainis. Not only are they an appreciated
foodstuff, but dates have strong, favourable associations with religion and health. For example, it is customary to eat one date
before breaking the Ramadan fast, and it is said that Mary fed dates
to Jesus (41). Such modern-day evidence for the special position of
dates strengthens arguments for a place for the date palm in ancient
religious traditions (42).
The fibrous parts of the date-palm have a very wide range of
uses, including roofing, matting and basketry, and one might expect
to find traces of this in the phytolith record. The numerous fragments of palm charcoal show that it was used for fuel.
The date stones give few clues as to how the date fruits were
used. The obvious way is as a dried or fresh fruit, and there are
numerous modern recipes, for example for dates pounded into a
smooth paste and made into biscuits, and for various combinations
EARLY
DILMUN DIET ET SAAR
of dates and flour. Dishes such as that described one hundred
years ago by Theodore Bent: "They make slzeerr~hfor their own
consumption out of dates dried for three days; then date-juice is
poured over them and sesame seeds, walnuts, or ginger powder
mixed with them" (43) are similar to those made today in Bahraini
homes. For example a dish called dabbus is made from dates mixed
with a little date syrup, with sesame and candied ginger on top.
This is an early morning food to eat with coffee.
An important modern culinary use, and one archaeologically
attested at other sites since c. 1750 BC, is as date syrup "dibs". In
Bahrain today, dibs (also known as "black honey" el nsal a1 aswad)
is much used as a sweetener with staple foods such as rice and
bread, as well as in confectionary.
Dates at their driest, most mature stage of ripeness ("tamur"),
arc packed in date palm leaf baskets in date presses ("madbasa")
in August-September. As the dates ferment, date syrup runs off.
Rougeulle (44) suggests that the word "diSpu", used from Old
Akkadian times onwards and conventionally translated as "honey"
(45), may have meant "dibs", prior to the advent of bee-keeping in
the first millennium BC.
Ancient "madbasa" are known from mid-second millennium BC
Qal'at el-Bahrain (46), 13th century AD Islamic levels at the same
site, and from 13th-14th century A D fort Sohar in Oman. An
undated, but presumably medieval madbasa, is in the courtyard of
the fort at Arad, Bahrain. All these have characteristic plastered
corrugated floors, with channels leading to tanks or earthenware
jars (47). Similar structures, but with flat plastered floors, have been
found at levels in Failaka dating to c. 1750 BC. Presses need not
have the characteristic corrugated floors of Gulf examples. In Iraq
today, dates are simply heaped in unroofed mud-walled enclosures
out in the date gardens (48). Occasionally the floor is covered in
hard pitch. The crucial identifier for date presses must be some kind
of sunken jar or tank to catch the liquid date syrup. Such structures
have not yet been found at Saar, but date syrup production may
have taken place in specialised areas of the settlement that remain
to be discovered.
A different kind of date syrup ("merees") is also produced in
Bahrain on a smaller, kitchen, scale by soaking dates in water and
squeezing them in cloth.
The part of the date that survives uneaten, and is therefore most
likely to be charred, is the stone, and it has a number of uses that
may be archaeologically visible. Today, date stones are ground into
small pieces (but not powder) and fed to animals mixed with wheat
bran. Whole stones can also be fed to animals after cooking in
water. This custom is described by a visitor to Bahrain in the late
nineteenth century: "Green dates (salang) are given to the donkeys
MARK NESBITT
for fodder, and to this the Bahreini attribute their exceptionally
good breed" (49). With a fat content of about 8%,and 5% protein,
date stones are a useful fodder (50). The absence of the perianths
that would normally accompany date stones suggests that the heavily fragmented charred stones at Saar may, at least in part, derive
trom debris from pounding. While stone fragments might be charred after passing through animals and being burnt as a constituent
of dung, the absence of charred dung fragments (especially as compared to the ubiquitous charcoal) suggests dung was not an important fuel at Saar. Even in the recent past, despite the apparent barrenness of the island, palm wood and shrubs have always provided
enough fuel to allow all dung to be used as fertiliser rather than
fuel.
In the light of evidence at Saar of bronze working, the observation of Strabo (16.1.13) (51) that in Mesopotamia: "The bronzesmiths use the stones of the fruit [of date] instead of charcoal" may
be relevant. Dowson similarly records that in modern Iraq date
stones were sold to charcoal makers, and after conversion into
charcoal, favoured as a fuel by silversmiths (52).
- --
0
05mm
F1g. 4.
Canary grass grain. Saar 1991.
L18:31:01.
Wild plants
The four wild grass seeds (Table 6; Fig. 4) have the extreme lateral
compression and convex dorsal and ventral profiles of canary grass
(Phalaris sp.). Canary grass is not common on the island today, but
R minor has been recorded in a variety of habitats. In Iraq this
species grows in depressions in the desert and is a weed of cereal
fields and irrigation canal banks.
The three Trifoliae seeds belong to one of the small seeded
members of the pea family related to clover. Such species grow
both in the desert and in irrigated fields. The weed seeds could
therefore be derived either from grain cleaning or from fodder, and
are not diagnostic of agricultural practices.
Indeterminate 1nateria1
A category used to define material that is clearly not seed or
charcoal. The large quantity of material in Q20-69/70 appears to
consist of tangled masses of cells, but awaits SEM examination.
Fuelwoods
Most of the samples contain some date palm charcoal, easily identifiable by the fibrovascular bundles embedded in a rather amorphous
looking matrix of cellular tissue. However, the rest of the charcoal
presumably represents other woody species of the island. While
EARLY DILMUN DIET ET SAAR
7iible 6. Cereal grains and weeds in the 1990 & 1991 Saar samples.
Cereals
Straight
Twisted
Indet.
hulled
hulled
Naked
hulled
barley
wheat
barley
barley
SAAR 1990
G17:29:01
G17:31:01
G17:32:01
G17:34:01
I
-
1
1
I
-
-
Weeds
Indet.
cereal
grain
Canary
Small
grass Leguminosae
Other
weed
-
-
-
-
-
-
-
-
3
-
-
-
23
33
4
3
3
1
1
SAAR 1991
EI6: lO:OY
E16:14:04
E17:02:24
F 16:25:03
M13:27:01
HI3:28:01
113:52:01
K18:19-20
L17:45-46
L17:70:01
L18:31:01
Q20:27:01
Q20:32:01
Q20:52:03
Q20:60:03
Q20:69:01
QZO:78:O.3
R20:20:03
Total
7
9
Onc hexaplo~dwheat rachis segment.
9
i-
One barley rachis segment
there are no truly wild trees on the island, except for the mangrove
swamps of the coast, there are plenty of shrubs, including the
dominant Zygophyllum gatcirense. Analysis of the charcoal (by Dr
Rowena Gale) will indicate whether the vegetation was substantially different 4000 years ago.
Agricultural systems on Bahrain
The main factor affecting agriculture on Bahrain is the availability
of water. Annual rainfall is highly variable, falling from November
to March and averaging about 6 cm/year, but with variation within
a decade from 14 cm to 0.4 cm (53).
Cultivation has therefore been restricted to flat land that can be
irrigated by gravity flow irrigation systems from natural springs
(54). Such land is restricted to a belt around the northern side of
the island, amounting to about 85 km2in all, less than 14% of the
MARK NESBITT
island's total land area (55). In the 1940s crops of wheat were
successfully grown on unirrigated land in thc valley of Rifat, but
yields were so variable from year to year that the enterprise failed.
Temperature, ranging from a low of 9°C in December to 44°C
in June is much less limiting, and a very wide range of fruits, cereals
and pulses can be grown successfully.
Hansen's description (56) of agricultural life in Saar in 1960 is
typical of village life on Bahrain in the recent past: the date gardens
were irrigated by four qanats and one shaduf. Some Indian Zebu
cows were kept permanently inside cowsheds in the compounds,
fcd on luccrne, datc stones and dried fish. A small number of fattailed sheep, duck and poultry were also kept in the village. Herds
of goat lived on plants in the desert surrounding the village. The
basic foods were rice, dates and fish, with other meats more rarely.
Even though oil had already affected life on the island, the husbandry of thc datc palnis was still a key activity.
Details of date cultivation are the same as elsewhere in the
Gulf (57). Somewhere between 500,000 (Popenoe's estimate in the
1920s) and 427,000 (Ahmed, mid 1970s) palms grow in date
gardens concentrated in a coastal belt around the northern half of
the island, where water from springs and qanats is available. Trees
are artificially pollinated in February and March, and dates appear
on the market from late July to late October. At present Bahraini
dates are only sold fresh, but this may be a result of declining yields
in recent years reducing the number of dates available for sale (58).
At least 15 varieties of date in Bahrain were described to me. The
varieties differ primarily in colour, time of harvest and uses.
Date gardens form the basic agricultural unit of Bahrain today.
A network of small channels takes water to rows of date trees and
to small rectangular areas of soil, surrounded by heaped up banks,
between the trees. There are three main elements to the date
gardens - the date trees, the fodder plants, and the other food
plants. This system may be as old as the cultivation of dates, as
once large-scale date cultivation started, with its requirement for
systematic irrigation, use of the shady, easily watered spaces between trees ivould naturally follow. At ancient Saar the ubiquity of
dates suggests that they were an important foodstuff, and largescale date cultivation must have been taking place in order to supply
sufficient to a town as large as Saar.
Today the list of food plants grown in the date gardens is still
close to those recorded early this century by Dowson (59) in Iraq:
onions, leeks, hibiscus, eggplant, beans, cucumbers, vines, ziziphus,
peaches, melons, figs and pomegranates. Given the restricted area
available for agriculture, production naturally focuses on foodstuffs
best eaten fresh, rather than on cereals such as wheat and rice which
can easily be imported from other countries. However, Albuquerque
EARLY DlLMUN DIET ET SAAR
described the island in 1510 as "noted for its large breeding of
horses, its barley crops, and the variety of its fruits ..." (60). Fodder
plants are also best grown locally as being too bulky to import.
Today handsome onions seem to be the most common of these
crops. In the third millennium, Sumerian written texts refer to "sumdilmunki", translated as Dilmun onions or garlic (61), although as
with Dilmun dates this does not necessarily mean that the onions
were imported from Dilmun.
The narrow range of food plants recovered at Saar does not
include any of these vegetables or fruits. The absence of onion seed
is not surprising as vegetables eaten green are unlikely to leave
many seed remains, but the lack of fruit is odd. The cherry-sized,
orange fruits of sidr are widely used in Bahrain today. Similar fruits
I o Y on
~u,
are found both on the wild species, Zizijd~us~ I M ~ ~ ~ I ~ ~ Mand
the cultivated Z. syim-christi. The large woody stone should char
well, and possibly survive uncharred, but careful checking of heavy
residues (and the extensive dry-screening programme) has not yet
recovered any. Ziziyhcrs wood (but not fruit) is present in Bronze
Age Failaka (62), and at Hili 8 in Oman at 2500 BC, and the fruit
stones have been recovered from fourth millennium layers at R135
in Oman (63). By the second millennium, major Near Eastern domesticatcs such as grape and pomegranate should have had time to
reach Bahrain. Again, if the archaeobotanical evidence is representative, this is further evidence for a conservative date-orientated diet.
Work on dental pathology has shown a very high occurrence of
caries at all periods on Bahrain, including the Middle Bronze Age
(skeletons dating from 2500 to 1700 BC, some From the Saar burial
ground). The nature of the caries points to a non-fibrok and tacky
food (64), and the pathologists who have carried out this work
plausibly suggest dates to be the major culprit - further evidence
of their importance in the Dilmun period. The details of the location
of the caries on teeth apparently point to a relatively minor role for
grain foods in the diet, and this complements the archaeobotanical
evidence of the much greater abundance of date remains.
The main fodder plant cultivated in the date gardens of Bahrain
today is lucerne, Medicago sul.ivu. Another fodder crop known from
similar current-day date gardens in the region is the clover bersirn,
Tr~fuliurndexarrd~inum,recorded in the AI 'Ain oasis (65). However,
we have no direct evidence for ancient fodder beyond the seeds of
canary grass and small Trifoliae, which could just as well be weeds
of cereal fields.
Keith Dobney's preliminary study of the animal bones from Saar
finds fish to be the main economic base, with some gazelle as the
other important wild animal, and sheep and goat, with some cattle,
as the main domesticates. These relative quantities (with the exception of gazelle, extinct in the wild in Bahrain) closely match Han-
MARK NESBITT
sen's twentieth century record for Saar village, and it is tempting
to speculate that herding and feeding patterns may have been
similar.
Bahrain's identification with the textually known trading entrepot
known as "Dilmun" and its role in the export of copper from Oman
to Mesopotamia is generally accepted (66). Much less is known
about the import or export of foodstuffs from Bahrain. Today,
virtually all rice and other cereals are imported, and there is no
reason why barley and wheat could not have been imported in
similar fashion in the past. As we have seen, the Dilrnun dates of
Mesopotamia were probably locally grown, and there is no literary
evidence for date trade in either direction.
An early 2nd millennium text from Ur mentions sesame oil "for
expedition to Dilmun to buy copper there" (67), and there is mention of issue of a barley ration to "Dilmunites" - who may have
been living at Ur (68).One Old Akkadian text (late 3rd millennium)
text reports a consignment of flour on a boat to Dilmun. This is
the sum of hard evidence for trade in foodstuffs with Dilmun.
The most striking aspect of food production in Bahrain is its
continuity. The diet of Saar some 4000 years ago - lots of dates,
some cereals, lots of fish and a little goat, sheep and cattle - can
be compared to that observed by Hansen in the village that is
the modern successor to ancient Saar, in 1960, and to Nieuhoff's
observations:
The common people make use of dates instead of bread or rice;
for it is observable that the ordinary food of the Indians [i.e.
natives] all along the coast from Basora to Sinde, is dates and
fish dried in the air; the heads and guts of the fishes they mix
with date-stones and boil it altogether with a little salt water,
which they give at night to the cows after they come out of the
field where they meet with very little herbage" (69).
Larsen (70) has integrated evidence for settlement patterns, trade
and environment for the 6000 years of known occupation in Bahrain. The resulting synthesis elegantly demonstrates three peaks in
population, each coinciding with increased foreign trade (and, in
the first case, increased water supplies). The three peaks are in the
Barbar I and I1 periods (c. 2100-1700 BC), the Neo-Assyrian and
Neo-Babylonian periods of the first millennium BC, and the
medieval period (14th century AD onwards).
Putting the Barbar I1 period, the time of Saar, in perspective,
Larsen writes of the early third millennium: 'Higher relative sea
levels may have reinforced artesian springs to provide more abundant water supplies ... This hypothetical increase in hydrostatic
pressure was synchronous with a period of increased rainfall that
watered the colluvial soils surrounding the Bahrain dome making
them available for additional cultivation', leading in the Barbar I
EARLY DILMUN DIET ET SAAR
phase to 'greater social differentiation, population growth, and increased demands on the agricultural system. Grains and other foods
were imported into the islands ... Dates from the various gardens
provided a durable export crop'. In the Barbar I1 phase, 'Increasing
agricultural demands also reached a maximum ... Settlements spread
to their farthest extent along the southwest coast' (71).
This reconstruction, which sets Saar firmly in the context of
expanding trade and increased water supplies (especially rainfall) is
attractive, but D. T. Potts (72) has drawn attention to some flaws
in the evidence. The securest part of Larsen's work is the archaeological survey: there is a peak of occupation (admittedly partly represented by the highly visible burial mounds) in the Barbar I and I1
periods, with 26 known sites, compared to just four in the following
Kassite period. Even though survey data of this type are open to
all kinds of biases, such a large difference is surely significant. The
difficulties lie in the explanations for this expansion. With regard
to climatic changes, as Potts points out (73), even the major shifts
between dry and wet climatic phases of the Upper Pleistocene and
Early Holocene are still the subject of widely diverging views, and
therefore the use of what are even less clear-cut data, the location
of 'lakeside' sites in one oasis in eastern Arabia (74) to indicate
higher rainfall in the third millennium is highly debatable. The
suggestion that rainfed cultivation of cereals extended over the area
of 68 km2 identified as colluvial soils hinges on this evidence for
increased rainfall. For trade, it is true that there is evidence of
extensive trading contacts, but evidence for trade in foodstuffs is
slim. As Potts points out, not only is there no evidence for the
export of dates then or at any other time prior to the late medieval
period, but the suggested export of dates to southern Mesopotamia,
a major area of date production, is particularly unlikely (75).
Looking at longer term patterns of change, Larsen focuses on
evidence for export of dates (as we have seen, not attested until
the 19th century) and on textual records from the Hellenistic and
medieval periods which, because they do not explicitly mention
cereal crops, are interpreted as evidence that they were not grown,
and must have been imported (76). However, in the 12th century
surplus production of barley on Bahrain is recorded, and in the
19th century wheat and barley are amongst the main agricultural
products of Bahrain (77). It is not surprising that some descriptions
of Bahrain mention the date gardens and not the cereals, as the
date palms are the most striking agricultural feature. While there
probably have been fluctuations in the area of cereals cultivated,
and while increased rainfall would allow a greater area of land to
be cultivated, the textual and archaeological evidence available at
present does not allow even tentative conclusions to be drawn on
these subjects.
MARK NESBI'I'T
In contrast, literary evidence for different types of crops does
exist, for example in observations made by one of Alexander the
Great's admirals, Androsthenes, around 323 BC and recorded by
Theophrastus. The text includes descriptions of mangrove and tamarind trees, and then discusses the cotton plant: 'They say the island
also produces the "wool-bearing" tree in abundance. This has a leaf
like that of the vine, but small, and bears no fruit, but the vessel in
which the "wool" is contained is as large as a spring apple, and
closed, but when it is ripe, it unfolds and puts forth the "wool", of
which they weave their fabrics, some of which are cheap and some
very expensive' (78). Pliny paraphrases this text, with some extra
details that indicate that cotton was still grown in the first century
BC. In the 14th century AD Ibn Batutta records of Bahrain, 'In this
place are palm enclosures, and pomegranates, lemons, and cotton
are cultivated'. Cotton can only have been grown in the date
gardens, whose area has always been limited by the availability of
water, and its importance as a cash crop through extensive stretches
of time, together with evidence for the abundance of cereals grown
on the island in the medieval periods, suggest that even in the
medieval period, a time of high population, Bahrain had the capacity
to feed itself without importing foodstuffs. However, it would be
unwise to extrapolate this productive capacity back to the second
millennium, as it may well have been dependent on an expansion
in the area of irrigated land owing to the introduction of the qanaf
system. This is most likely to have reached Bahrain in the first
millennium BC.
In summary, the basic elements of diet observed in Bahrain in
the recent past seem to have been established by the early Dilmun
period - dates, fish, cereals (wheat and barley then, rice today),
domestic animals. Two changes that have occurred are in the range
of fruits grown on the island and, perhaps owing to the introduction
of the new irrigation technique of the qanaf, in the growing of cash
crops such as cotton at some periods in the island's history. Flotation of samples from sites covering these later periods could be
highly informative.
Ancient agriculture, Bahrain and the Gulf
Few plant remains are available for comparison in the northern Gulf:
Failaka (79), Qal'at al Bahrain (80) and Tell Abraq (81) have all
produced abundant date remains, and little more can be said. The
elucidation of the earlier history of the date palm awaits the recovery of plant remains from earlier sites. Much more comparative
material is available from sites to the south. The most obvious
difference in contemporary assemblages from these southern sites
EARLY DILMUN DIET ET SAAR
is the presence of sorghum, both in the mountains of Yemen and
in the oasis of A1 'Ain.
It has often been suggested that Arabia, with its extensive trading
contacts and central position, acted as a bridge for the spread of
crop plants such as sorghum from Africa to India (82). Sorghum is
known to have been domesticated in Africa, as this is where the
wild ancestor grows (83). The earliest archaeobotanical records from
Arabia come from the third millennium DC. At Ra's al-Hamra on
the coast of Oman, sorghum has been found in layer 4 at the site
RH5, radiocarbon dated to about 3000 BC (84) and at Hili 8, a
settlement mound in the AI Ain oasis in Abu Dhabi, some 500 km
SE of Bahrain. Charred seeds and mudbrick imprints from phases I
and I1 (radiocarbon dated to about 3000 BC) include wheat, 2-row
barley, sorghum, jujube, melon, date stones and palm wood (85).
Sorghum has also been found at Wadi YanZ'im (radiocarbon dated
to c. 1750 BC) in Yemen (86). Sorghum has now been identified
from a number of early second millennium sites in India (87), and
it does seem likely that crop plants were moving between Africa,
Arabia and India from at least the third millennium.
It may seem surprising that early second millennium Saar, with
its trade links with Oman and, perhaps, the Indus valley does not
have sorghum. While it is difficult to be absolutely certain, owing
to the poor preservation of charred remains, that sorghum was not
grown at Saar and Failaka, there arc grounds for regarding the
northern and southern parts of the Gulf as separate agroecological
arcas. Today sorghum is mainly a crop of subSaharan Africa, Yemen, and lndia (88). Despite Dilmun's trading links with Oman and
the Indus valley, its location in the northern part of the Arabian
Gulf means that it n~ouldnot be surprising if its agricultural patterns
are more typical of Mesopotamia.
To what extent the presence of sorghum in oases in southern
Arabia in ancient times made the agricultural economy different to
that of Mesopotamia and the northern Gulf is at present an open
question. The introduction of highly productive, irrigated summer
crops similar to sorghum, such as millet, sesame and cotton, does
seem to be linked to major economic changes in other parts of the
Near East at other times, for example in the Iron Age, the Islamic
period (89), and in the early second millennium on the Indian subcontinent (90).
Future work
Unless unusually carbon-rich contexts are discovered, flotation in
future seasons is likely to recover much the same kind of material. A
limited flotation programme is justified because additional material,
especially of the cereals, would be useful, and as material accumu-
MARK NESBITT
Fig. 5.
Arabian Gulf sites mentioned
in the text.
such as ernmer wheat,
lates, the continuing absence of food
sorghum and Ziziphus would be more solid evidence of non-use.
While flotation has recovered sufficient charred plant remains to
give worthwhile information on foodplants (and fuelwoods) at the
site, fundamental questions such as whether the cereals were imported, and how important they were, remain unanswered.
Sites such as Saar that have poor preservation of charred material
are excellent candidates for other, more recently developed, techniques. The analysis of residues from 13 sherds from the 1991
season is in progress. The collection of seed imprints from sherds
has proved worthwhile at other Gulf sites, but a range of sherd
material was examined in the 1991 season and appears very unpromising. A larger number of the "vegetable-tempered type
would be worth inspection. Phytolith analysis identifies the silica
bodies of plants. These survive without any need for charring or
other special conditions, and at this site could provide a more
EARLY DILMUN DIET ET SAAR
representative record of plant use. Cereal grains and chaff contain
very distinctive sheets of phytoliths, and if chaff (and, therefore,
grain processing) is present at the site in large quantities this technique should detect it.
Acknowledgements
I am grateful to date growers and many other residents of Bahrain for sharing
their knowledge so freely. Conversations wit.h Mohammad and Abdul Rahman
Juma on date husbandry, Samira al-Saie and Miriam al-Hirrni on dates in cookery,
and Dr Philip Basson on botany were especially helpful. I also thank Delwen
Samuel for reading the manuscript, lane Goddard for drawing the seeds and Kate
Morton for the map. The Department of Antiquities kindly gave permission for
part of the material to be exported for analysis, and the National M~~seuln
of
Bahrain provided workspace.
The major sponsor of the London-Bahrain Archaeological Expedition is United
Energy International under its president, Mr. John ~amuels.The environmental
programme was also supported by a grant from the National Geographic Society.
The Expedition received additional support from the following companies and
institutions: Bahrain Petroleum Co. (BAPCO); Bahrain Telecommunications Co.
(BATELCO), British Academy, British Bank of the Middle East, British Council,
Bahrain Maritime and Mercantile International, Budget Rent-A-Car, Caltex (Bahrain), Cathay Pacific Airways, Computerised Technical Services, Diplomat Hotel,
H.M.G. Foreign and Commonwealth Office, Gredco-Ansari Consultants, Gulf Air,
International Aeradio Ltd, Manama Rotary Club, Mansouri-Mclnerney, McDonald
Institute for Archaeological Research, Mohammed Jalal & Sons, Philip Morris,
Royal Society of St. George, Society of Antiquaries and Yateem brothers.
References
1. Killick RR et al. London-Bahrain Archaeological Expedition:
1990 excavations at Saar. AAE 2: 1991: 107.
2. Williams D. Flotation at Siraf. Antiquity 47: 1973: 288.
3. Killick et al. London-Bahrain Archaeological Expedition: 133.
4. Popenoe PB. The dafe palrr~. Coconut Grove, Miami: Field Research Projects, 1973: 74.
5. Willcox G. The plant remains from Hellenistic and Bronze Age
levels at Failaka, Kuwait. A preliminary report. In: Calvert Y &
Salles J-F, eds. Failaka: 1986-1988. Lyon: Travaux de la Maison de
]'Orient, 16: 1990: 43.
6. Good illustration in Neef R. Plant remains from archaeological
sites in lowland Iraq: Hellenistic and Neobabylonian Larsa. In: Huot
J-L, ed. Larsa: truvaux de 11985. Paris: CNRS, Editions Recherche sur
les Civilisations, Mkmoire 83: 1989: 151.
7. Powell MA. Sumerian cereal crops. Bullefin on Sumerian Agriculture 1: 1984: 48.
8. Ellison R, Renfrew JM, Brothwell D & Seeley N. Some food
offerings from Ur, excavated by Sir Leonard Woolley, and previously unpublished. ]ounral of Archaeological Science 5: 1978: 167.
MARK NESBITT
9. Exceptionally favourable conditions of preservation - in burnt
rooms - has given direct evidence for the production of pearl
barley, in the form of pots of barley husks found in Iron Age levels
at the excavations of Cordion and Sardis in Turkey (author's work
in progress).
10. Zohary D & Hopf M. Donmtication of plnnts in the old world.
Oxford: Clarendon, 1988: 147-149.
11. Costantini L. Considerazioni su alcuni reperti di palma da dattero e sul centro di origine e l'area du coltivazione della Phoenix
dac!ylifeva L. In: Gnoli G & Lanciotti L, eds. OrienfaLa losephi Ztcci
Memoriae Dicafa. Rome: lstituto ltaliano per il Medio ed Estremo
Oriente, Serie Orientale Roma 56(1): 1985: 209.
12. Huot J-L. 'Ubaidian villages of lower Mesopotamia. In: Henrickson EF & Thuesen I, eds. Upon this foundatio~- the 'Ubaid reconsidered. Copenhagen: Museum Tusculaneum, 1988: 19-42.
13. Safer FM, Mustafa A & Lloyd S. Eridu. Baghdad: State Organisation of Antiquities, 1981.
14. Zohary D & Spiegel-Roy P. Beginnings of fruit growing in the
old world. Science 187: 1975: 319.
15. Hopf M. Jericho plant remains. In: Kenyon K & Holland TA,
eds. jericho. Volunw 5. The potfey phases of the Tell and other finds.
London: British School of ArchaeoIogy at Jerusalem, 1983: 576.
16. Zohary and Spiegel-Roy. Beginnings of fruit growing.
17. Costantini L. Ecology and farming of the protohistoric communities in the Central Yemeni Highlands. In: de Maigret A, ed.
The Bronze Age culture of Hawlan at-Tiyal and al-Hada. Rome: lstituto
Italiano per i1 Medio ed Estremo Oriente, in press: 187.
18. Potts DT. The Arabian Gulf in antiquity, l. From prehistory to the
full of the Achnemenid Empire. Oxford: Clarendon Press, 1990.
19. Ellison et al. Some food offerings from Ur.
20. Costantini, Ecology and farming of the protohistoric communities.
21. Costantini L & Biasini LC. Agriculture in Baluchistan between
the 7th and the 3rd millennium B.C. Newsletter of Baluchistan Studies
2: 1985: 16.
22. Rowley-Conwy P. Remains of date (Phoenix dnctylifern) from
Failaka, Kuwait. In: Hsjlund F, Danish ~rchmenlogicnlinvestigntions on
Failaka, Kuwait. The second n.rillenniurn settlements. 2 . The Bronze Age
pottery. Aarhus: JASC XVII(2): l98 7: 181.
23. Costantini, Ecology and farming of the protohistoric communities.
24. Costantini L & Biasini LC. Le piante di Yelkhi. In: Quarantelli
E, ed. Ln ferra tra i due Fiurrii. Venti anni di archeologia italiana in
Medio Oriente. La Mesopotnmia dei tesori. Torino: I1 Quadrante Edizioni, 1985: 57.
25. van Zeist W & Vynckier J. Palaeobotanical investigations of
EARLY DILMUN DIET ET SAAR
Tell ed-Der. In: de Meyer L, ed. Tell ed-Der II! Leuven: Uitgeveri]
Peeters, 1984: 119.
26. Potts, The Arabinn Gulf in antiquity, l.
27. Costantini & Costantini. Palaeoethnobotanical investigations in
the Middle East and Arabian peninsula, 1986.
28. Costantini, Considerazioni su alcuni reperti di p a l m da dattero:
214.
29. CAD. S. Volume 15, 1984: 373.
30. Postgate JN. Notes on fruit in the cuneiform sources. Bulletin
on Sumerian Agriculture 3: 1987: 115.
31. AHw: 1404.
32. Landsberger B. The date palm and ifs by-products ~iccordingto the
cuneiform sources. Graz: AfO, Reiheft 17, 1967.
33. CAD G. Volume 5, 1956: 65.
34. Landsberger, The dale palm and ifs by-producks. 56, n. 200.
35. Hartman LF & Oppenheim AL. On beer and brewing techniques
in ancient Mesopotamia. New Haven, Connecticut: JAOS, Supplement, 10: 1950.
36. Killick et ul. London-Bahrain Archaeological Expedition: 126.
37. CAD. A. Volume 1(1I), 1968: 338.
38. Potts, The Arabia11 G u y in anfiquity, I.
39. Kramer SN. Dilmun: quest for paradise. Antiquity 37: 1963: 111.
40. The barley grains from Saar, while too battered and too few
for measurement, are not noticeably small.
41. For this and culinary information I thank Samira al-Saie and
Miriam al-l-Iirmi of the National Museum of Bahrain.
42. cf. H~jlundF. Date honey production in Dilmun in the mid
2nd millennium B.C.: steps in the technological evolution of the
madbasa. Prlliorietlt 16(1): 1990: 77.
43. Bent JT. The Bahrein islands, in the Persian Gulf. PRGS 12:
1890: 1.
44. Rougeulle A. Des "etuves" a dattes a Bahrain et en Oman: le
probleme le I'apparition des techniques de transformation de la
datte. Pnliorient 8(2): 1982: 67. English translation published in Dilmun 13: 1985/86: 34.
45. CAD. D. Volume 3. 1959. 161.
46. Herjlund, Date honey production in Dilmun.
47. See Rougeullc, des "etuves" h dattes a Bahrain and Herjlund,
Date honey production in Dilmun for full descriptions.
48. Dowson VWH. Dates and date cullivation of lruq. Purf I. Cambridge: Heffer, 1921.
49. Bent, The Bahrein islands, in the Persian Gulf.
50. Winton AL & Winton KB. Structure and composition of foods.
Volume 11, Vegetables, legumes, nuts. New York: Wiley, 1935.
51. The Geography of Stmbo. Transl. H.L. Jones. Harvard: Loeb,
1941: 215.
MARK NESBITT
52. Dowson, Dates and date cultivation of lrnq: 60.
53. Good RD'O. The Bahrain islands and their desert flora. In:
Cloudsley-Thompson JL, ed. Biology of Desert. London: Institute of
Biology, 1954: 45.
54. Recent day irrigation systems have been based on qanafs, a
technique of water-distribution that seems to have been developed
in the first millennium bc (Potts, The Arubiun Gttlf in anfiquity, I).
Earlier irrigation must have used less complex gravity-fed systems,
perhaps combined with simple lifting devices. It is difficult to assess
the effects of the introduction of qannts on agricultural production
in Bahrain.
55. Larsen CE. Life and lund use on the Buhrain islands: thegeourchaeology of an ancient society. Chicago: University of Chicago, 1983.
56. Hansen HH. Invesfigations in a Shi'a uilluge in Bahrain. Copenhagen: National Museum of Denmark, Ethnographical Series 12,
1968.
5 7. cf. Popenoe, The date pulm, Ahrned R. The date palm in Bahrain.
El Budaiya': Ministry of Commerce, El Budaiya' Agricultural Experiment Station, Extension Unit Publication, 2: nd, C.'1977.
58. The decline in date production since the early part of this
century is discussed by Rumaihi MG. Brrhrain: social and political
change since the first World War. London: Bowker, 1976. Reasons for
the decline include land tenure systems, the existence of better paid
employment in the oil and service industries, and the disintegration
of the spring-fed irrigation systems.
59. Dowson, Dates and dafe cultivation of Iraq: 11.
60. Quoted by Bent, The Bahrein islands.
61. Potts, The Arabian Gulf in anfiquity, 1: 183.
62. Willcox, The plant remains from Hellenistic and Bronze Age
levels at Failaka.
63. Biagi P, Maggi M & Nisbet R. Excavations at the aceramic
coastal settlement of RH5 (Muscat, Sultanate of Oman) 1983-5. In:
Frifelt K & Ssrensen P, eds. South Asiun Archaeology 1985. London:
Curzon Press, Scandanavian Institute of Asian Studies, Occasional
Papers, 4: 1989: 1.
64. Littleton 1. & Frslich B. An analysis of dental pathology and
diet in historic Bahrain. Pallorient 15(2): 1989: 59.
65. Stevens HJ. Changing agricultural practice in an Arabian oasis:
a case study of the AI 'Ain oasis, Abu Dhabi. G] 136: 1970: 410.
66. Potts, The Arnbian Gulf in anfiquify,1.
67. Potts, The Arabian Gulf in arifiquity, I: 223.
68. Potts, The Arabia~Gulf in antiquity, l: 226.
69. 1662, quoted by Ouseley W. Travels in the various countries of
Lhe East, I. London, 1819: 228.
70. Larsen, Life mid land use on the Bahmin islands: 92-113, 200207.