Bahrains oldtidshovedstad gennem 4000 år


  • T.G. Bibby


Bahrain, prehistoric capital, oldtidshovedstad, forhistorisk hovedstad, Ras al-qala’a, kronologi, chronology, excavation method, udgravningsmetode, excavation of, udgravning af, pottery types, keramik typer


The Hundred-Meter Section

The "tell" of Ras al-Qala'a lies in the centre of the northern coast of Bahrain Island in the Persian Gulf, at the head of a shallow bay. It receives its name from the 'qala'a', or fort, which crowns the mound and which was occupied, and perhaps built, by the Portuguese in the early 16th century. Within the crumbled walls of this fort the Danish Archeological Bahrain-Expedition each year establishes its camp, and in the cool of the evening its members, taking the air upon the ramparts, have a bird's eye view of the holes in the ground in which their days are spent:

It is an impressive view from the ruined ramparts. Far away on the right, at the horn of the great bay, the low white houses of Moharraq stretch along the shore; closer in cluster the wind-towers and minarets of the capital Manama. On the left, beyond and above the acres of palm-tops, the coast of Arabia can be seen on the clearer days, stretching endlessly towards the north and, as the evening draws in, the red glow of the spill-gas flames of the Arabian oilfields begins to stain the heavens all along the western horizon.

The mound of Ras al-Qala'a is surrounded, on the three landward sides, by miles of date-palm plantations, and it is only by reason of the height of the mound above the surrounding countryside, and the height of the fort ramparts above the mound, that we can see the desert and the gardens and the township - and the clustering thousands of the gravemounds - beyond the palm-belt. But these date plantations terminate abruptly where the Qala'a mound begins. The mound is large, some 400 yards from north to south and twice that distance from east to west, but low by Mesopotamian standards, not more than 15 meters high. It is composed entirely of the ruins and debris of the considerable city which has at various periods occupied the site, and its comparatively modest height is due to the fact that this city was built of stone and not, as in Mesopotamia, of mud-brick. Stone buildings survive longer than buildings of mud-brick, and, whereas a brick building, once fallen, is levelled off and a new building erected above it, in the case of a stone building the materials are often salvaged, even foundation walls being frequently dug up, and the stone reused in the next building phase. Consequently a stone-built city climbs upwards on its own ruins much more slowly than one of more fragile materials. And, incidentally, the persistent quarrying for stone makes the identification and interpretation of occupational levels much more hazardous.

The interpretation of occupational levels is necessarily of the highest importance to discovering the history of a town site, particularly where, as in Bahrain, the history of the area, the sequence of event and cultures, is initially completely unknown. It is therefore one of the rules of the game that, as soon as the first soundings have shown that a site does in fact contain early material, an extensive trench should be dug, going down to bedrock or to virgin soil. This trench should be dug in such a way that everything found in each layer of occupational debris is kept separate, and keyed in to a drawing of the trench wall which will show every change in structure or composition of the soil, which denotes a change in the type of habitation of the site. Such a drawing of a vertical section through the mound will be capable of giving the complete history of the portions of the site through which it runs. Here can be seen protruding walls and the lines of stamped earth or clay floors, marking phases of building. Above the floors will lie the stratified debris and dark earth accumulated during settled occupation. Above these again will lie the jumbled stone and mortar left when the buildings fell to ruin or were pulled down, with perhaps the thick charcoal layer that denotes destruction by fire. This may be followed by a new building phase, or by the drifts of windblown sand which tell of a site left desolate. In the section will be seen, too, the pits and trenches dug for storage, for graves or for stone-quarrying, but in general the layers will lie roughly horizontally, and it will of course always be true, that the lower levels must precede, chronologically, the levels above.

The objects recovered from each of these separate layers, bones and beads, worked stone and metal, and above all potsherds, fall similarly into a chronological series, and can give a picture, from bottom to top, of the vicissitudes of the site. A gradual development from one pottery type to another will tell of long, settled and comparatively undisturbed occupation, whereas an abrupt change in pottery and artifacts will denote a gap in occupation or the supersession, probably by no means peaceful, of one culture by another. It is among these collections, too, that one may hope to find the imported objects which may, by comparison with already known and dated cultures elsewhere, be able to put an absolute date to a level and illustrate a cultural connection with some other region.

It is important, however, to realise the limitations of such a section trench. In the nature of things it cannot be of very great width, and therefore the horizontal area uncovered is limited. Consequently it will rarely uncover a complete building, or provide a broad view of any one particular phase of occupation. For this it will be necessary to supplement the section trench by large-scale area digging, often by widening the initial trench at points of particular interest.

The procedure above outlined has been consistently followed at Ras al-Qala'a. The original sounding 1) showed the site to have been occupied at least as early as the middle of the First Millennium BC, and revealed a building of unusual interest and massive construction. This original sounding has since been widened by an extensive 'area-excavation' together with two small soundings down to bedrock which showed that the occupation of the site went back probably to the Third Millennium BC 2). In 1955 a start was accordingly made on the driving of a section trench in from the northern slope of the city mound on the very edge of the sea. The trench ran due south and its commencing point was chosen at a place where recent high tides had washed away part of the mound, exposing two blocks of worked limestone.

There are several reasons for commencing a section trench on the edge, rather than in the middle, of a city mound. The mound is normally lower there and there is less earth to move, while such earth as is moved is easier to dispose of. But chiefly, such a trench should at an early stage reveal what type of fortification wall, if any, the city possessed, and such fortifications, interesting enough in themselves, can be followed fairly easily to a gateway. Gateways were important places in early eastern city life and can be expected to produce quantities of dateable material. The only disadvantage of digging inwards from the outskirts of a mound is that, if the city has expanded gradually from an inner nucleus, the section trench will not for some time reach the earliest periods of occupation. We are far from certain that we have yet reached the earliest city at the Qala'a, even after three years' work.

Expecting fortifications, we were hardly surprised to find that the two limestone blocks already exposed proved to be facing blocks of a well-built wall, and the sharp curvature of this wall was soon explained when a more massive wall, 2.5 meters thick, was discovered behind it, and the first wall was seen to belong to a turret abutting on the wall (Fig. 1). Pottery recovered from the occupation layer of the turret already showed a considerable quantity of multicoloured glazed ware and some sherds of the green celadon china characteristic of the Sung Dynasty of China (960-1250 AD).

The main defensive wall stood solidly on bedrock to a height of 3 meters. It was of unshaped stone set in gypsum cement, a type of concrete, and had originally been faced with large and finely cut blocks of limestone from the island of Jida off the Bahrain coast. Most of these facing stones had been removed, only the lowest courses remaining, and many stones of the type of those surviving could be seen incorporated in the walls of the 'Portuguese' fort further inland.

Inside the wall, but at a level with the top of the surviving portion, the section trench uncovered the floors and lower parts of the walls of finely constructed buildings. The pottery found in association with these floors was of the same type as that from the turret floor and will be discussed in more detail later.

On digging below this level it could be seen that the buildings had been very solidly constructed, the walls resting on foundations which stretched 3 meters down to rest on bedrock. But cut into by these foundation walls were the walls of a previous building phase, of a completely different plan and orientation from that above. The pottery from this earlier phase, too, was completely different from that above, being characterized by thin-walled shallow bowls either of a gray-green glazed ware, or else with a burnished red or black wash.

As the trench progressed towards the south it became clear that it was following, on the upper building level, a street running inwards from the turret, and bordered on both sides by substantial buildings. 18 meters from the defensive wall this street debouched into a paved square. And when the trench was widened to investigate this feature it became clear that this square formed the central point of a completely symmetrical complex of streets and buildings. The square was exactly quadrilateral, with its sides running precisely north­south and east-west, and it sloped inwards from all sides to a circular drain in the middle. From the centre of each side a short street led off in each of the four cardinal directions. The east and west streets were not followed; but those to the north and the south were practically blocked, after only 7 meters length, by a room built out from the east side of the street, leaving only a narrow alley which, in the case of the north street, led up to the fortification wall at the point where the turret was situated (Fig. 2).

In the course of the 1955 campaign the southern alley was not followed in its full length. The rooms which occupied the quadrants between the streets were just as geometrically and symmetrically constructed as the streets. They showed signs of varying use. One, which lay so close to the present surface that floor and walls had only partially survived, had an earthenware drainpipe beneath its floor which led, cutting through a stone threshold of the previous building phase, to a seepage tank below the street outside. The next room had a gypsum cement floor with two cement-lined pits symmetrically placed at each end. Two of the rooms were equipped for the collection of date juice; the walls were · coated with clay plaster and the floor, also of clay, consisted of baulks and channels running the length of the room and sloping gently to a tank, in one case of cement, in the other clay-lined, lying below floor level at the lower end of the room. Rooms of this nature are still in use for the same purpose, the baskets of dates being piled up at the upper end of the room, and their own weight gradually pressing out the juice of the fruit, which runs down the channels to the tank.

The reason for abandonment of this dwelling area was not apparent. In one of the rooms immediately south of the fortification wall there was considerable evidence of fire, a thick layer of ash, and large quantities of the collapsed roof, which had been made of alternate layers of matting and bitumen, a much more weather-proof ceiling construction than that used at the present day in Bahrain. But while fire had clearly destroyed this room, the destruction had been very localised, and there was no evidence of violent destruction of the remainder. On the contrary, there was some evidence of a gradual deterioration of the strict, almost military, ordering of the quarter. Post holes in the north street leading from the square showed that a temporary building, probably one of the local palm-leaf huts, had been at some time erected there, apparently, to judge by the number of quernstones lying in and around it, a mill-wright's shop. And in a corner of the square itself the blackened stones of a hearth suggested a period when even the square itself was taken into use as a camping-place.

Our belief that we had found the fortification wall, albeit a late one, of the city, and were digging the ruins of the city, clearly Islamic, which lay behind it, received a rude shock when, at the beginning of the next season's work in 1956 we continued the line of the section towards the south. Almost at once a new fortification wall was struck, at the end of the south alley and precisely symmetrical with the northern fortification wall at the end of the north alley. When a corresponding turret was found to the south of this southern wall it became clear that we had not been digging in the city at all, but in an isolated fort, or rather a fortified residence - presumably a palace - between the city proper and the seashore. Further digging along the line of the southern wall in that and the following season identified two corner towers and enabled the presumed plan of the fort to be worked out in some detail (Fig. 3).

The section trench had now been driven some 59 meters southward from the seashore, and the prehistoric city had not yet been reached. It was now continued onward, in a width of 5 meters.

Within the fortified palace it had only proved possible to make two small trial sondages down to bedrock along the line of the section, as it was not desired to destroy the later buildings. But south of the fort, and at the same Islamic level, an unbuilt area stretched for some distance, and here it was possible to carry the section trench down to bedrock over a stretch of 8 meters (63-71 meters South). In the whole of this length no remains of buildings were found, except for one wall, running obliquely across the excavation. There were also, in the upper levels, three lslamic burials which were left undisturbed. The 5 meters depth of soil was divided horizontally, in accordance with changes in colour and texture of the soil, into 20 different levels, and all potsherds and the artifacts from each of these levels were kept separate and later analysed and compared. Details will be given later, but at this point it may be said that two complete breaks in the types of pottery found occurred, below the 6th level, and below the 16th. Levels 1-6 contained the lslamic pottery already familiar from the fortified palace, mainly characterized by polychrome glazed ware and green Chinese celadon. Level 7-16 formed a deposit 2.2-2.5 meters deep, with very homogeneous pottery, the most common types of which were small thin­walled bowls covered with grey glaze or with red or black wash, frequently burnished (Figs. 4-5). In level 7 a small terracotta female head (Fig. 6) was found, and in level 10 a portion of a rouletted black-on-red Attic bowl. Levels 17-20 again showed a completely new list of pottery types. The number both of sherds and of types was small and many of the sherds were wave-washed, not unnaturally considering that the final layers 19-20 comprised the sand of the original beach immediately overlying bedrock and less than a meter above mean sea level. These levels were characterized by a preponderance of thick honey­coloured or caramel-coloured sherds, including a number of narrow bases and pedestals (Fig. 7).

At this point it seemed as though a beginning had been made towards establishing a historical sequence. Three successive "cultures", each clearly separated from the others, had been identified, and there seemed good possibilities of dating fairly closely the upper two of them. But the city for which we were looking seemed as far off as ever. At the period of the "lslamic-palace" levels the area just dug had been open ground surrounding the fortified palace. At the time of the "glazed-bowl" levels there had perhaps been scattered walls and houses here, but nothing that could be called a city. And at the time of the "caramel­ware" levels the area had been foreshore and nothing else. And yet the initial sondage in the centre of the city had shown substantial buildings whose levels were characterized by red-ridged pottery of a type which appeared nowhere in the three cultures which here overlay bedrock, but which was found in the Barbar temples in a context which suggested a Third Millennium BC date 3).

The 1956 season was drawing to a close, and the fasting month of Ramadan was approaching. It was decided to jump 25 meters and sink a final sondage on the line of the section, 99-104 meters from its commencement at high-water mark. It soon became clear that this pit had completely different characteristics from those previously dug along the section. Here 19 levels were distinguished.

Levels 1-6 contained mainly the pottery characteristic of the "Islamic-palace" phase, but amongst it were smaller quantities of the typical pottery of the "glazed-bowl" and "caramel-ware" phases, and even - at the two extremes - sherds of the red-ridged "Barbar" ware of the Third Millennium BC and of the blue-and-white Chinese Ming porcelain which characterizes the Portuguese occupation of the 16th century AD. This mixture was not difficult to account for. The section had at this stage reached the immediate vicinity of the moat of the 'Portuguese' fort, and the upper levels here were clearly composed of upcast earth from the digging of this moat.

With level 7 the contents of the layers changed abruptly, but not, as might have been expected, to those of the "glazed-bowl" phase, but to the typical "caramel-ware". The "glazed-bowl" phase proved, in fact, to be entirely missing here.

"Caramel-ware" persisted until level 10, but with level 11 the pottery type changed again, and here, at last, came the typical red-ridged "Barbar" ware in large quantities. It will be described later, but its main characteristics were unmistakeable (Fig. 8). Almost 95 % of all sherds were of thin brick-red grit-tempered ware, and over 75 % of these bore the unmistakeable "Barbar" ridges, while the typical "Barbar" rims occurred in large numbers. Together with the uppermost "Barbar" layers came the first building level, a corner of a substantial house of stone construction with a threshold entrance (Fig. 9). As the levels went down the pottery gradually changed character. "Chain-ridged" ware, in which the parallel horizontal ridges were not even but were regularly depressed to form a chainlike pattern, appeared and gradually increased in proportion to the normal ridged ware until it became, by level 16, the only type of ridged ware found (Fig. 10). At the same time the proportion of red ware dropped, and brown and buff ware comprised up to 40 % of all sherds. It was clear that the "chain-ridged" phase develops gradually into the true "Barbar" phase, unlike the abrupt transitions between the other phases identified. The boundary between the two phases is thus indeterminate. At level 17 a new building phase was found with a stone wall running parallel to the main section, containing a doorway besides which lay a hollowed hinge-stone (Fig. 9).

The very great difference between the cultural content of the portion of the section dug to bedrock at 63-71 meters from high-water mark and that dug at 99-104 meters suggested strongly that between the two lay the boundary of the "Barbar" period city, probably a fortification wall. Consequently, at the beginning of the 1957 season the section trench was continued backwards from the last hole dug for a distance of 14 meters (85-99 m S).

Considerably more traces of buildings were found in this excavation than previously in the area south of the Islamic palace. Just below the surface (level 1) lay a cement water channel running the length of the excavation and clearly contemporary with, or later than, the Portuguese occupation of the fort. Half a meter further down (level 5), in the eastern side of the excavation only, lay the floors and stumps of walls of buildings dated, by the pottery found at their floor levels, to the period of the Islamic palace. To the west these buildings ended abruptly, cut away, as it appeared, by the edge of a large pit dug in Portuguese times apparently to obtain building stone. Immediately below this level the defensive wall of the "Barbar-period" city began to appear, a massive work of gypsum-concrete, 3.5 meters thick, running obliquely across the excavation from ESE to WNW .. Extension of the area excavated to both sides along the line of the wall showed that, to the east, the remains of the wall survived almost to the surface and had been incorporated into the buildings of the "lslamic-palace" period. To the west the pit dug in Portuguese times went ever deeper, and ended by taking a considerable bite out of the fortification wall itself.

While the general direction and core construction of the wall were clear, it had been exposed - at least on the outer, northern face - to such a degree of weathering by nature, quarrying by man and alteration to suit later building phases that it was not easy to gain an idea of its original appearance. It did not go down to bedrock, but its inner face, consisting of a dry-stone wall, descended to within 120 cms. of bedrock, 80 cms. lower than its outer face. This would suggest - though only a section through the fortifications can confirm - that the dry-stone portion of the ramparts forms the earliest phase, and that the concrete outer face was added later, at a time when the ground level, both within and without the wall, had risen considerably.

Outside - to the north of - the wall the first 85 cms. of "lslamic-palace" period remains were immediately succeeded by 1.8 meters of "glazed-bowl" levels. These overlay the floor level of a building phase characterized, by walls abutting onto the outer face of the city wall, which had clearly already at that period lost its original facing stones. One of these walls contained a doorway with a cut stone threshold of exactly the same type as that discovered in the "glazed-bowl" phase below the lslamic palace. A meter above this floor level a pointer was given to the age of these levels by the discovery of a sherd of coarse straw-coloured pottery with, inscribed upon it, part of a name in Greek characters, -  παις (Fig. 11). Below the "glazed bowl" levels, and still 70 cms. above the foot of the wall, pure "Barbar" levels commenced, and these continued, developing gradually into the "chain-ridged" levels, until bedrock was reached.

Within - to the south of - the wall, the succession of levels was entirely different. Below the Islamic house lay a sterile layer of green sand 10 cms. thick, and below this lay, not more than 2 meters below the surface, levels containing the red-ridged sherds of the pure "Barbar" period. The sterile layer sealing the "Barbar" levels was broken through time and again. A shallow excavation of the "caramel-ware" period dipped down to rob part of the wall forming the upper "Barbar" building level, before continuing deeper to the south in the thick levels found the previous year. Through the edge of the sloping "caramel-ware" excavation a grave had been dug, containing one of the pitch-covered "bath-tub" coffins of the same type as those previously discovered in the central area of the Qala'a city 4). The coffin contained an undisturbed skeleton but no grave goods (Fig. 12). The type can, however, be dated fairly accurately by the contents of those previously discovered to the 7th century BC. These two excavations were in their turn dug into by the quarrying depredations of the "lslamic palace" and the "Portuguese" periods, with the final result that, although the layer of sterile sand in fact sealed over half the area excavated, it did not appear at all on the main section, and only for a distance of 20 cms. on one of the cross-sections (at 98 m. S).

Below this sterile level, and the pits dug through it, the "Barbar" and "chain-ridged" levels stretched 4 meters down to bedrock. They contained the two building levels already attested from the previous year and, in addition, a large number of important "small-finds". Of these undoubtedly the most significant are three steatite stamp seals of the type known from discoveries in Mesopotamia and in Mohenjo-daro in the lndus valley (Fig. 13), and already attested from Bahrain in a surface discovery 5). These will be discussed in a later section. One (Fig. 13 a) lay in the floor deposit to the upper building level (level 14). The second (Fig. 13 b) is without dating significance, as it lay in a secondary deposit, the "fill" of the robber trench which had removed part of the inner facing stones of the fortification wall. The third (Fig. 13 c) lay in a layer of beach sand (level 21) which, soon after the erection of the fortification wall, had been laid immediately behind it - presumably to fill up irregularities in ground level. Seal c is thus clearly considerably earlier than a. The layer of beach sand was of interest in other respects. It contained a portion of the rim of a stone bowl ornamented with engraved circles with centre points. Portions of similarly decorated stone bowls were also found in the layer immediately below (22) and the layer next adjacent above (19). A stone bowl with similar decoration, in the possession of the Bahrain government and found close to the finding place of the surface-discovered stamp seal, has been described in earlier reports 6). The layer of beach sand also contained, in addition to two beads, a very large number of small scraps of oxydized copper. Other discoveries of significance were, in the same level (14) as the upper stamp seal, a piece of ivory, clearly a piece broken off an unworked tusk where the sawcut removing the tusk had overlapped itself; and in the level below the lower stamp seal (23) a rimsherd bearing a Sumerian record of capacity. This short inscription is illustrated and discussed on p. 165 by Dr. Læssøe, and here it need only be said that the style of script is not inconsistant with Early Dynastic or Sargonid date (roughly 3000-2000 BC).

In these levels there was also a small quantity of painted pottery (Figs. 14-15) in general of much finer paste and from smaller vessels than the red grit-tempered ridged ware among which it lay. The designs were in all cases geometric, cross-hatched lozenges, zigzags, triangles and horizontal lines. Apart from a few specimens in the upper levels of red zigzag lines on buff, apparently applied after firing, all the painted decoration was in black, mainly over a reddish brown slip, though occasionally on a buff and once on a greenish background.

This is the evidence to hand at the completion of the 1957 season of excavation, and it is clear that we have sufficient to gain a fairly clear idea of what has been happening, at least on this tiny portion of the northern beaches of Bahrain, and particularly when it has been happening. We have identified six successive cultural epochs in the history of this peripheral area of the city at the Qala'a. It would already perhaps be in order to give these epochs the Roman numerals, starting from the bottom, which mark the final crystalization of a site's history, but, as our area is still peripheral, and there may come further surprises as we work inwards, it will be best to keep to our descriptive names, the names which have grown up naturally in the course of tossing baskets of potsherds from one archeologist to another.

We have then, in chronological order: -

a)                         the "chain-ridge" period,

b)                         the "Barbar" period,

c)                         the "caramel-ware" period,

d)                         the "glazed-bowl" period,

e)                         the "lslamic-palace" period, and

f)                          the "Portuguese" period.


Apart from the first two, which develop imperceptibly from the one to the other, each of these periods is clearly distinguished from every other in style and type of pottery, while many of the periods include a number of "small-finds". There is thus a good chance of being able to ascribe a fairly accurate date to some, if not all, of the periods, and thus to be able to reconstruct in some measure the absolute history of this part of Bahrain. This will involve going in some detail through the characteristics of each of the six periods and drawing parallels, where this is possible, with periods of known date elsewhere in the world. That sort of thing is only interesting if one is interested in that sort of thing, and the reader who prefers results to methods is urged to jump to page 162 where the conclusions will be found painlessly tabulated.

The "chain-ridge" period (Fig. 10):

In this period the pottery is predominantly (50-60 % of all sherds) of thin, grit-tempered, dark-red to biscuit ware. A greyish-white slip is occasionally found. The dominant type is a large globular jar either terminating in a simple, slightly thickened, incurving rim, or in a thick shoulder, a narrow vertical neck 3--5 cms. high and a rolled rim. The body, though never the neck, of the jar is often ornamented with applied horizontal ridges either of the "chain" pattern described earlier or, more rarely, of continuous sharp-edged form. The jars have either round or slightly flattened bases, but the continuation of the horizontal ridges right down to the foot of the jar frequently converts the lowest ridge into what is in effect a ring base. All pots are wheel-made.

Other characteristics common both to this and to the following period are described following the latter.

The "Barbar" period (Fig. 8):

Here the preponderance of thin, red, grit-tempered ware is even more marked, ranging from 75 to over 90 % of all sherds, and of these over three-quarters are decorated with the typical horizontal ridges. These are, however, not now sharpedged or of the "chain" pattern, but are round-topped; they are no longer applied to the pot but are built up from the paste by finger-pressure. A cream slip is occasionally added. The large globular shape, with or without neck, is still overwhelmingly dominant, but the rims now show greater variation and more careful fashioning. Where the vertical neck is found it now terminates in an outturned rim of triangular section 7); where the jar terminates in the incurved rim this rim is considerably thickened and either rounded, concavely flattened, or narrowed again to a slightly upturned edge. Several of these vessels have, in addition, a short wide spout (about 2 cms. wide and 1.5 cms. high) immediately below the rim.

Amongst the preponderating mass of the globular pottery of these two periods there is a small quantity of other pottery of characteristic type. Perhaps most easily recognised are small shallow bowls with a flat base and a straight outward-sloping wall terminating in a very short vertical wall and a slightly outturned rim. These occur in buff and straw-coloured, straw-tempered ware, and do not exceed 9 cms. in height and 15 cms. in diameter.

From both periods very large vats, up to a meter in height and in diameter, are found. From the first period these are of thick green-yellow, straw-tempered and poorly fired clay with applied horizontal ridges at wide intervals, and broad flat rims. From the "Barbar" period they are of better fired red clay, with deep closeset applied ridges continuing all the way down to the base where the final ridge forms a little ring base surrounding a bung-hole.

Two rare but characteristic rim types of the two periods are a completely rectangular rim, and a flat sharply-outturned rim on an incurving shoulder, forming a wide "collar"-rim.

The painted pottery found from these two periods has already been described (Figs. 14---15). It was considerably better made, of a finer paste and apparently on a faster wheel, than the ware hitherto described.

The dating of these two periods has not been easy. The objects found in association with "Barbar" ware at the name-site of the Barbar temples strongly suggested that this period should be attributed to Early Dynastic IIIb ("Royal Graves" - 2459-2304 BC) or at latest to the following Akkad dynasty (2303-2108 BC) 3). Unfortunately the pottery of these periods discovered in Mesopotamia has not been extensively published, and such publications as do exist, entirely from sites in Sumer, appear to give no points of resemblance with the typical Bahrain ware of these periods. It was therefore of very great interest to discover that the Oriental lnstitute of Chicago, which is making a survey for the lraqi Government of village sites of all periods in the Diyala Basin area of what was formerly Akkad, has worked out a sequence of the most common types of pottery in use in Akkad at all periods from Ubaid to Early Babylonian/Kassite, for use in dating surface indications 8). This sequence is based largely on unpublished material from the excavations of the lraq Government Antiquities Department at Tell al-Dair. Photographs of the pottery types used as a criterion of the Akkad dynasty in this survey resemble in all respects the pottery of the "Barbar" period of Bahrain, the ridged ware and the short wide spouts being especially unmistakeable. Pending comparison of actual type-sherds, and bearing in mind that this equation was already likely on other grounds, it would appear reasonable to ascribe the "Barbar" period to the Dynasty of Akkad (2303-2108 BC) and the "chain­ridge" period from which it developed to (with considerably greater reservation) Early Dynastic IIIb (2459-2304 BC).

It might be thought that the "small finds" from the "Barbar" period, in particular the three steatite stamp-seals (Fig. 13), would be of assistance in dating this period. In fact they are of very limited help. This type of seal is well known, in particular as a result of an article by Professor C. J. Gadd 9), who first pointed out that the type (as well as other types containing lndian subjects, and inscriptions in the lndus Valley script) occurred both in Mesopotamia and in the cities of the lndus Valley civilization. These seals are, in fact, now regarded as the principal evidence for trade between Sumeria and the Indus Valley, and for the date of the lndus Valley cities 10). Disregarding here the seals of other types with lndus affinities found in Mesopotamia, we have now a total of 24 known specimens of the round steatite seals of the type shown here. Of these 3 have been -found in Mobenjo-daro on the lndus 11), 4 in Bahrain, and 17 have come, or probably come, from Mesopotamia, including 13 from Ur. It is necessary to point out that they are not the shape of seal typical of the levels in which they occur either in Mesopotamia, where the typical shape is cylindrical, or in the lndus cities, where it is square. On the other hand, no other shape of seal than this has been found in the "Barbar" levels of Bahrain.

The dating of the seals has. hitherto been beset with difficulties. The Mohenjo-daro seals are not intrinsically dateable - in fact they form the main evidence for the date of the levels in which they lie, rather than the reverse. As Gadd had pointed out, only 3 of the Mesopotamian seals published are accompanied by any evidence of date, and this evidence is of a very loose character. One (Gadd 12) is described as found in an "upper rub­bish, Kassite? level" above a Larsa house; its stratigraphy has been stated to be "very doubtful" 12). Another (Gadd 15) is described as of Sargonid date, and the third (Gadd 16) as found in the filling of an Ur II tomb shaft, though Frankfort would prefer an Akkadian date 13). It would thus appear that it can only with certainty be said of the Mesopotamian seals that they are not recorded as found earlier than Akkad/Ur II (about 2300 BC) or later than the Kassite period (1762-1165 BC). And by implication Mohenjo-daro and the lndus civilization must be given the same possible range of date. It would seem, then, that, if the pottery equation "Barbar" period = Dynasty of Akkad can be regarded as sound, it can contribute more to the dating of the stamp-seals, and thereby of the whole of the lndus civilization, than the stamp-seals can contribute to the dating of the "Barbar" period.

A type of pottery technically closely resembling "Barbar" ware, and like it bearing horizontal ridges by finger pressure, is of frequent occurrance in the graves of Cemetery H at Harappa in the lndus valley 14). The Harappan vases possess, however, much taller and more curved necks than the pure "Barbar" type.

The "caramel-ware" period (Fig. 7):

For this period the chief characteristic is the very large number of pointed and pedestalled bases in thick, coarse, honey or caramel-coloured, straw-tempered ware. These bases are extremely numerous, amounting to 5-10 % of all sherds found. No complete or completeable vessels have been obtained from this period, but some at least of the pedestal bases, which can he of all heights up to 8 cms., form the supports for very small, almost flat upper surfaces, scarcely wider than the pedestal is high. The pointed bases are at least as numerous as the pedestals and are noticably thick and solid. No small finds have come to light from this period, but pottery of precisely this appearance and colour is illustrated by the Diyala Basin Archeological Survey as typical in Akkad of the Early Babylonian/Kassite period (1894-1165 BC) 8).

The "glazed-bowl" period (Figs. 4 and 5):

Undoubtedly the characteristic of this period which most springs to the eye is the very large number of small bowls of thin, straw-tempered ware. These bowls vary very greatly in size, shape, treatment and decoration. They fall into three groups, glazed, washed, and untreated. The glazed bowls show the greatest variation, having simple rims, simple incurving rims, outward-rolled rims, or vertical sides ending in three horizontal corrugations. The glaze is normally gray and crackled, though occasionally a grass-green or a golden, iridescent glaze appears. The washed bowls bear either simple rims or rims expanded into a cylindrical form with a sharp inward-turned edge. They have either a red wash on a buff ground or a black wash on gray. The wash is frequently burnished, sometimes in stripes, and very frequently the wash is confined to the interior of the bowl and the top 1.5 cms. or so of the outside. The untreated bowls are normally buff in colour, with simple, or simple in­curving, rims. Their flat bases are frequently oddly polygonal rather than round. All bowl types are occasionally found with a base consisting of three equally-spaced lugs.

Few pottery types other than the bowls have been identified. One common type, however, is a round-bodied glazed vase with a narrow neck, a massive rim and two pierced-lug handles (Fig. 5).

The dating of this period is difficult, largely owing to the dearth of publications on early glazed ware. The glazed ware of somewhat similar type found at Taxila, Dura Europos and Antioch-on-the-Orontes is generally ascribed to the period 300 BC - 300 AD, but on these sites the simple bowl types are regarded as an early feature, suggesting that our period may commence earlier than these sites 15). Red and black wash bowls are normally explained as imitations of Greek, or even Roman, ware 18), but they have been found in 7th-6th century BC contexts at Tell Beit Mirsin in Palestine 17) and it is even possible that the Roman terra sigillata was rather copied from Asiatic originals.

There can, in any case, be little doubt that the black-on-red rouletted Attic sherd, found in the upper third of the "glazed-bowl' levels, belongs to the first half of the third century BC, or perhaps a little earlier 18). The terracotta head (Fig. 7) from the uppermost level of this period, with its Hellenistic, almost Parthian affinities, can best be ascribed to the commencement of the Christian Era.

In general it would appear safest to ascribe a date of about 500-0 BC to this period, with the possibility that it may extend a century or so on either side.

The "lslamic-palace« period (Figs. 16, 17 and 18):

This period is characterized by a number of distinctive pottery types:

1)                         large narrow jars (up to a meter high) of buff ware with rims sharply inturned.

2)                         amphorae with high necks supported by three double (or occasionally triple) handles (Fig. 16).

3)                         large wide jars with rounded shoulders and short neck, the shoulders ornamented with three horizontal bands of multiple incised lines, the centre band of the three being undulated (Fig. 17).

4)                         "coffee-pots", conical pots with two tall handles, the one pierced to serve also as a spout. These vessels bear geometric (particularly "ladder") ornamentation in dark red paint over a pink wash (Fig. 16).

5)                         "soup-plates", glazed either in whole colour deep yellow or brown, or with patterns in black normally radiating outwards from the centre in four equidistant and symme­trical patterns (Fig. 17).

6)                         quantities of olive-green celadon Chinese ware, often with embossed leaf decoration and with ring bases.

Only the latter two types are approximately dateable, the geometrically patterned plates being a characteristic of lraqi ware of the 9th and 10th centuries AD19), while export of olive-green celadon commences with the Sung Dynasty of China (960-1250 AD). As the embossed leaf patterns on celadon are an early feature, the "lslamic-palace" period should most probably be dated to the 10th century AD, with a permissible error of perhaps a century to either side.

The "Portuguese" period:

The characteristic pottery of this period is still unknown, as no settlement levels from this period have been excavated. But its remains cover the surface and are found in quarrying trenches wherever we dig. The period is characterised first and foremost by the presence of the famous blue-and-white porcelain of the Ming Dynasty of China (1368-1644 AD), pottery which is found spread over the Eastern world from the Philippines to West Africa in colossal quantities and often in association with the forts and trading stations of the Portuguese, who in the early 16th century as the first of the European nations broke into Asiatic trade. There can be little doubt that this blue-and-white ware (with its Persian imitations) is contemporary with the Portuguese occupation of Bahrain in the 16th and early 17th centuries.

With all the reservations which we have indicated in the above description and discussion, we can therefore now repeat our list of periods, and this time add the rough absolute dates to which they can be ascribed.

a)                         the "chain-ridge" period          - possibly Early Dynastic III b, 2459-2304 BC.

b)                         the "Barbar" period                                                - probably Akkad Dynasty, 2303-2108 BC.

c)            the "caramel-ware" period                  - possibly some short time within the range of the Early

                                                                                   Babylonian/Kassite periods, 1894-1165 BC.

d)                         the "glazed-bowl" period        - probably Achaemenid-Seleucid period, about 500-0 BC.

e)                         the "Islamic-palace" period    - probably about 900-1000 AD.

f)                          the "Portuguese" period         - about 1500-1650 AD.

It would be of interest finally to attempt a reconstruction, in the broadest outlines, of the history of the northern edge of the Qala'a city on the basis of this rough dating and of the other evidence given by the section trench.

The story opens some 2500 BC with the earliest building levels of the city of the "chain-ridge" peoples. The city was well-built, of stone, with mathematically laid-out streets and houses aligned by the cardinal points of the compass - evidence of town­planning more resembling that of the Indus valley than of contemporary Sumer. Yet the inhabitants were using measuring bowls inscribed in Sumerian, and presumably imported from the north. The city was apparently unfortified (unless an earlier wall lies hidden within the later ramparts), but ended abruptly a hundred yards from high-water mark. This hundred yards of foreshore served without a doubt as the harbour of the city, and on it would be drawn up the shallow-draught vessels which plied up and down the Gulf, calling in for water and for trade at the northern coast of Bahrain. The seals, beads and fragments of stoneware and pottery found in the deposit of beach sand behind the later wall suggest that the foreshore was a busy place, the scene of unloading and loading, and perhaps marketing, merchandise from the whole of the known world. And the considerable quantity of scraps of copper in the sand gives a line on one of the main articles of trade, and perhaps can even be taken as evidence of ship-building on the beach.

The last of the unfortified levels of the city ends in conflagration, in a thick black burnt level and floors scattered with blackened storage jars with carbonised content (which will, incidentally, in the course of time, enable this important phase of Bahrain's history to be dated by the Carbon 14 process). The city appears to have been destroyed by fire, and it would not be impossible to equate the date with the recorded conquests of Sargon of Akkad about 2300 BC. There is no break in the cultural sequence, however, the same artifacts and pottery being used by the inhabitants after the destruction as before. But when the city is rebuilt, on the same plan, a massive stone rampart is erected around it. From then on, for an indefinite period, successive habitation levels within the city wall bespeak settled and continous habitation, and gradually raise the ground level within the walls. Stamp seals still testify to cultural connections with the Indus valley, and a scrap of sawn ivory to trade in this raw material. When and how the end of this period of habitation came is unknown. There is some evidence that the area of the city covered by the section trench has been levelled off by later inhabitants, whereby traces of the final phase of the "Barbar" period city may have been erased.

However that may be, by the Kassite period, some time in the Second Millennium BC, this area of the city was no longer inhabited. The only evidence of the "caramel-ware" people consists of quarrying pits dug to salvage building stone from the lower levels. But the fact that these pits were filled up with obvious kitchen-rubbish, including a large quantity of pottery, suggests that their habitations were not far away, probably elsewhere on the tell. It is not impossible that this clear drop in the size and importance of the city is to be seen in relation to the conquest of the Indus valley by the Aryans and the cessation of trade with India.

Even as late as the seventh century BC this part of the site was still uninhabited, but the bath-tub coffins of the burials of the period again suggest that there was a settlement at that period elsewhere on the extensive tell.

Shortly afterwards, however, perhaps in the fifth century BC, a new building phase commences in the area of the section trench. There is little or no building on the higher ground within the ruined walls of the then two-thousand-year-old city, but the hundred yards of foreshore are covered with scattered buildings, many abutting on the outer face of the ancient ramparts or even dug into the thickness of the wall. Again, trade appears to have lain at the root of the increased prosperity of the city. Sherds of Attic ware prove that trade goods from Greece were reaching the Gulf in Achaemenid or Seleucid times, while the Greek name scratched on a potsherd suggests that not merely Greek goods but Greek seamen and perhaps Greek ships were frequenting the Gulf.

With the breakup of the Seleucid Empire it would appear that the city once more stagnated, and there is no further trace of occupation in our area until the tenth century AD. Then we find houses built once more within and upon the ancient city wall, while, at a safe distance from them on the edge of the sea, lay the fortified palace of the ruler. It was probably short-lived; there is only one building phase, and it has been suggested that this period coincides with the rule in Bahrain of the Carmathian heretics in the century preceding 1058 AD.

After the defeat of the Carmathians no more buildings are erected in the tiny area of the tell of Qala'at al-Bahrain here dealt with. The Portuguese period of the sixteenth century AD, like the "caramel-ware" period of approximately the same date BC, is represented only by quarrying pits. But the quarrying pits are immense. To build the fort which towers above the tell they ruthlessly rip out the wall-blocks of the Islamic palace, tear up the thresholds of the "glazed-bowl" people, hack their way deep into the mighty defense wall of the "Barbar" city. The archeologist should without a doubt preserve an objective attitude to the peoples of all periods of history and prehistory But I find it difficult to love the Portuguese …

T. G. Bibby





Bibby, T. (1957). Bahrains oldtidshovedstad gennem 4000 år. Kuml, 7(7), 128–163. Hentet fra