- 1 For overviews of the project, see Keall 1983; 1994; 1999a; 1999b.
1This paper is based upon the results of surface reconnaissance survey and excavation by the Canadian Archaeological Mission of the Royal Ontario Museum (CAMROM) on the Red Sea Tihâma coast (see Fig. 1). The Mission’s mandate is to investigate the city of Zabîd and to explore its hinterland1.
2The main aim of the Zabîd Project is to try to determine how the city evolved from the time of its reported foundation in the 9th century: what was the basis of its economy that allowed sponsors to build citadels, mosques and madrasas? Who were its leading figures and most dynamic rulers? Who were the intellectuals, and what was the measure of its material culture? What was its relationship to the outside world? One might also hope to pass judgment on what was the quality of life for the ordinary people. Besides that there are some very basic urban issues like how was water delivered to the city; the effectiveness of its sanitary systems; the quality of its dwellings and industrial goods, and the nature of its consumer items. In other words, everything that makes for the definition of an impressive city. The 15th century historian of Zabîd, Ibn al-Dayba‘ who was a native of the city, is a little guilty of hyperbole in its description, but his words give some sense of Zabîd’s former repute2: “It is the mother of cities and the centre of all kinds of arts, towards which scientists from all around the world travel to acquire from its knowledge.” The relationship of Zabîd to its port sites has been of major interest since the inception of the Project.
- 3 al-Mad‘aj, 1988, p. 9, n. 88.
3During the long history of Zabîd, several ports on the Red Sea were associated with the medieval city. Among them, Ghalâfiqa purports to have been Zabîd’s first port. Some modern historians have suggested that it was from here that fifty tribesmen of the al‑Ashâ‘ir set sail, along with Abû Mûsa al-Ash‘arî, in the year AH 7/628 AD, as a delegation to meet the Prophet in al-Madîna3. When the Prophet delivered his call to the Yemeni tribes to join the cause of Islam after the conquest of Mecca in AH 8/629 AD, the two agents he sent to Yemen were the aforementioned Abû Mûsa al-Ash‘arî and Mu‘âdh b. Jabal al-An/sârî. They, too, are likely to have arrived in Yemen by way of the port of Ghalâfiqa. But while one may visit today on the Tihâma coast a place called Ghulayfiqa (sic), with a natural shallow anchorage for small fishing boats, and observe pottery on the ground from the 9th- late 11th centuries, there is no tangible archaeological trace of remains of a port that has been discovered there from the first two centuries of Islam; nor do the early Islamic writers specify that it was actually Ghalâfiqa whence the delegation set sail and whither they returned.
4As will be explained below, this does not automatically deny the role of Ghalâfiqa as Zabîd’s first port. But the notion has been repeated ad infinitum by modern commentators to the point that it is difficult to identify the earliest genuine citation for Ghalâfiqa being the earliest port for Zabîd. It is after all entirely feasible to reach Jidda by a land route through the /Hijâz, by way of Najrân. It is from later writers that we may extrapolate the idea that Ghalâfiqa was Zabîd’s first port, and that it was from here that the famous delegation to meet the Prophet “set sail”.
- 4 For a recent review of the role of /Husayn Ibn Salâma, see Peli, 2008.
5One of the important early references to Ghalâfiqa appears in the Ta’rîkh al-Yaman, a text composed by ‘Umâra al-Yamanî before his execution in AH 569/1174 AD. Here we read about the impressive accomplishments of the late Ziyâdid wazîr /Husayn Ibn Salâma who put into place infrastructure to support the Hajj pilgrimage traffic through Yemen, around 1000-1010 AD4.
- 5 ‘Umâra al-Yamanī, p. 11.
6Along the coastal road in the CAMROM study area ‘Umâra cites amongst others the caravan stations of al-Khawkha, al-Ahwâb and al- Ghalâfiqa5. But the factual reality is that this work was being conducted over two centuries after Abû Mûsa’s epic mission to meet the Prophet, and we must be cautious about the validity of these statements without critical concern for ‘Umâra’s sources. However, ‘Umâra’s description of the coastal road invites us to consider Ghalâfiqa’s history in relationship to that of other ports accessible from Zabîd.
7As part of CAMROM’s aim to document features in the hinterland of Zabîd, earlier work had targeted the area to the south of modern Ghulayfiqa, in the bay of al-Fâzza (see Fig.2), which is another name associated with a port that had served Zabîd in medieval times. Confusingly, the medieval writers speak of al-Ahwâb, al-Fâzza, and al-Buq‘a to denote coastal sites located in the same area6. The most problematic is the latter, which in Arabic simply connotes “place”, and which therefore could be anywhere. For instance, on the 1:50,000 topographic maps for Yemen there are several references to “al-Buq‘a” marked along sections of the coast to the south of al-Fâzza, so that it is risky to suggest a specific identification for this place. Yet, based on archaeological realities, CAMROM’s fieldwork in 1997 resulted in a hypothetical resolution of the conflicting evidence.
8CAMROM’s firm conclusion is that the elusive site of al-Ahwâb is to be identified with archaeological remains visible on the surface at the southern end of the al-Fâzza bay. The fishing village of al-Mutayna is the nearest modern settlement, lying slightly inland, sheltered from the sea-winds by groves of wild dûm-palms. The village of al-Fâzza itself is much further north, in the central part of the bay.
- 7 See Ciuk & Keall, 1996, pls. 30-31.
9The settlement remains of al-Ahwâb lie on top of an old stabilized sand-dune, covered now with potsherds that are red from their iron-rich clay content, which explains the local toponym Kitf al-A/hmar (literally = “red shoulder”). Test trenches sunk in 1997 on top of the sand dune revealed no traces of any tangible structures. Only the pottery remained as evidence of a former settlement, one likely built of palm-fronds, in the manner of the fishing village huts of today. The pottery is limited in time range to the 9th-10th centuries. It includes the distinctive turquoise-glazed storage jars made in Iraq that are ubiquitous around the Indian Ocean in early Islamic times. The site assemblage also includes types of the classic wavy-lined decorated household wares that were produced in Zabîd after its 9th century foundation7. A cautionary note is necessary here, to the effect that CAMROM has documented this pottery as having been made according to the Zabîd standard of function and design, but made with local clay, as has been observed at Ghalâfiqa as well (see below).
10On the site of Kitf al-A/hmar itself, the palm fronds of the original dwellings have disintegrated, leaving no trace. Other soft materials from the occupation have also been gradually blown away by the strong on-shore winds, leaving only pottery (and some glass) on the deflated surface. On the flat ground on the seaward side of the dune there were recorded four small structures of mud-brick, one of which was a mosque (see Fig. 2). The modest nature of these structures, and the absence of any other substantial buildings on the site, reveals that al-Ahwâb was not a grand port settlement in the mode of a Sîrâf or a /Suhâr where merchants lived in grandiose high-rise dwellings and sustained rich lifestyles. Conceivably its modesty means that al-Ahwâb only served as a port of entry during the very limited monsoon season. However, one must be open to the idea also that it simply was only a very modest port.
- 8 Cf. Rougeulle, 2005, p. 226.
11The proportion of imported foreign items in the ceramic surface record is very small, compared to that of other contemporary Indian Ocean port sites, such as Sharma8. Yet a huge, formally laid out cemetery to the immediate south of the Kitf al-A/hmar settlement, with which it is assumed to be contemporary, implies that al-Ahwâb functioned more than just as a modest fishing village. A thousand or more preserved graves speak of a sizeable population. The cemetery was laid out on the top of a large stabilized barchan dune that has resisted erosion until recently, simply because it had been covered by another barchan dune. The graveyard is notable for the consistent practice at the time of burial of covering each grave with a layer of beach pebbles, and keeping those in place by means of a border of clay. Today the clay retains more moisture than the surrounding sand, and in the first moments of daylight before the sun has dried the surface dew, the clay shows up for a moment as a positive outline of the grave.
12There are also a few isolated graves on the flat ground to the seaward side of Kitf al-A/hmar, not far from the small mosque. The grave coverings are distinctly different from those in the main cemetery, being composed of conch shells and lumps of coral, rather than pebbles. Yet their slightly elevated position relative to the deflated ground around them suggests that they are old. But whether they are contemporary with the life of the port-site is a moot point. Conceivably they are to be associated with the later period (Rasûlid era) use of the area described here.
- 9 al-Khazrajî, I, p. 133, 288.
13On the inland side of Kitf al-A/hmar, where run-off was trapped to furnish wet clay, there are the remnants of a brickyard, of which some over-fired examples remain unused and unusable in their firing stacks. Their dimensions show that they were made specifically for the building of a substantial structure to the site’s immediate west. The walls of that structure have been almost entirely systematically mined later for re-cycling of the bricks, as is the pattern of archaeological remains in all of the Tihâma. But single period pottery fragments recovered from the building site allow one to attribute the structure to Rasûlid times, and one may assume that this is the famous “beach-house” or “sea-villa” where the Rasûlid sultans resorted to from time to time, as frequently noted by al-Khazrajî9. Incidentally, visitors to the beach residence entered by way of al-Fâzza. The settlement of al-Ahwâb must have been deserted by then.
14The port-site that functioned during this era in the al-Fâzza bay lies inland from the modern village of that name, which now lies at the end of what is now a newly engineered paved highway from the market town of Suwayq. For some years in the 1980-1990s the archaeological site was flagged by the presence of a magnificent dûm-palm that marked a saint’s grave (waliyya). Graves of this kind in the Tihâma are often placed on ground where there has been settlement in the past. Curiously, seemingly in keeping with the common Tihâma practice of placing oil lamps or candles on saint’s graves, in the hopes of receiving the intercession of the saint, this grave was covered with modern light bulbs, of both the incandescent and fluorescent kinds.
- 10 See Ciuk & Keall, 1996, pl. 37.
15Adjacent to the dûm-palm in the 1980s were visible relics of a past settlement, in the form of remnants of storage vessels preserved in situ to around one third of their original height. These storage vessels, in the so-called “Trackware” tradition, can be dated to around 11th-13th centuries10. A few potsherds were also recovered from the surface that can be dated to the 13th-14th centuries. But pottery of the 16th century was entirely absent. Apart from the settlement itself, and the later saint’s grave, there was another vast cemetery contemporary with the settlement site and formally laid out precisely in the same manner with pebble-covered graves as the one south of al-Ahwâb, implying once again a sizeable population. It, too, could be documented by CAMROM in 1997 because of the movement off-site of a large barchan dune that had covered the cemetery in the 1980s. It can be observed that, once subjected to erosion by the strong on-shore winds, sites like this are badly deflated already after a decade of exposure. We may propose that the archaeological remnants adjacent to the dûm–palm / waliyya site represent the port settlement of al-Fâzza in the Rasûlid era.
- 11 Cf. Keall, 2001b, p. 222-223.
16The third port settlement in the bay of al-Fâzza is next to the al-Fâzza mosque. It is an iconic feature for the entire area. Stylistically the mosque is likely to have been built no later than the middle of the 16th century, for the zone of transition used to transfer the square shape of the ground plan to the round shape of the dome, employs a squinch. This squinch (hooded niche) technique was generally phased out in the Tihâma by the middle of the 16th century through the adoption of a pendentive, comprising superimposed layers of bricks laid in a saw-tooth pattern11.
- 12 Cf. Keall, 2001a, p. 42-43, pls. 4 & 5.
17Sadly the sea is in the process of undermining the al-Fâzza mosque, as the soft sediments it is built upon are eroded away by relentless wave action on the shore. Twenty-five years ago the shore was fifty meters or more distant. Now the waves lap at the base of the mosque. Pottery on the surface of the ground next to the mosque can be dated to the 16th century Ottoman occupation of Yemen, and to more recent times, through identification of ceramic materials produced in the workshops of Hays12. Again, most significantly, there is an extensive cemetery adjacent to the mosque, to its east, that by far exceeds in extent the needs of the inhabitants of the handful of today’s fishing huts. In other words, it seems reasonable to claim that the mosque site provides the evidence for the third of the al-Fâzza bay ports settlements. It is worthy of note that, as with other medieval Islamic cemeteries in the coastal reaches of the Tihâma where significant graves often sport headstones that are recycled from Late Prehistory, some of the graves in the al-Fâzza cemetery have headstones of basaltic pillar fragments, likely from nearby al-Midamman where in the 2nd millennium BC they had been set up as menhirs.
- 13 ‘Umâra al-Yamanî, p. 4, n. 5.
- 14 Ibn /Hâtim, I, p. 248; Vallet, 2006, p. 330.
- 15 Ibn al-Dayba‘, p. 49.
18For clues as to the naming of this settlement site and its large cemetery, one may turn to Kay’s footnotes to ‘Umâra’s text where Kay refers to al-Khazrajî’s description of the city of Zabîd13. It was a round city, he cites, with four gates of which one (the western one) is called Bâb al-Nakhl (for the date-palm plantations to the west). But, according to al-Khazrajî, writing in the middle of the 14th century, it used to be called Bâb Ghalâfiqa whence the road led to Ghalâfiqa and al-Ahwâb. The last author to refer exclusively to the west gate as Bâb Ghalâfiqa is Ibn /Hâtim, who died in 130514. Ghalâfiqa is cited as having been once the port of Zabîd, but it fell into decay and was superseded by al-Ahwâb, which (at the time of al-Khazrajî’s writing) was called al-Buq‘a. Writing at the end of the 15th century, Ibn al-Dayba‘ has the same observation15. The name of al-Buq‘a, then, can be associated with the last phase of major settlement in the bay of al-Fâzza.
19The erosion of the headlands on which the mosque of al-Fâzza was built, and the movement of the port sites, serves to illustrate how much the coastline changes in the Tihâma as the result of seasonal sea-currents, tides and storm waves. As headlands are eroded away, sandbars are built up towards the north. This action creates a bay that can serve as a sheltered anchorage, the forerunner of a port. As the wave action continues, and the sandbar extends further, the original open but sheltered bay collects sand as the current drops. It is argued here that this is the natural phenomenon that caused al-Ahwâb to be abandoned, because the harbour had filled with sand. Today, between Kitf al-A/hmar and the al-Fâzza bay is a treacherous salt-flat (sabkha) that likely reflects the former harbour. The port settlement was then moved further north, to al-Fâzza, nearer to what was then the anchorage. With time the anchorage moved yet again, to al-Buq‘a, at the northern end of today’s bay.
- 16 Niebuhr, 1968, p. 323.
- 17 Niebuhr, 1969, pls. 7 & 8.
20The explanation of this natural process offers the chance to make sense of the archaeological remains that have been recorded at Ghalâfiqa. In 1763 Karsten Niebuhr spoke disparagingly of the run-down conditions of the village, as well as the difficulties of reaching the port because of the coral reefs16. Chances are that the coral reefs were always there, even in Ghalâfiqa’s heyday as a port. In all likelihood, small craft would have been needed to ferry people and goods from shore to larger vessels that stood in deeper water outside of the reefs. What had probably changed well before Niebuhr’s time was that sandbars had made the sheltered anchorage for small craft no longer viable. Sandbars in the vicinity of Ghulayfiqa today extend now well beyond the original bay (cf. Fig.1). But a reflection of Ghalâfiqa’s former grandeur is to be seen in the two Kufic-inscribed tombstones that Niebuhr documented during his visit17.
- 18 For an overview of the architecture of 6th century Yemen, see Costa, 1992 and Finster, 1996.
- 19 On Pre-Islamic Ghalâfiqa, see also Beeston, 1988, p. 1-6.
21Close to the modern village of Ghulayfiqa is a large enclosure, with a mi/hrâb in the qibla wall, indicating its role as an open-air mosque (mu/sallâ). The enclosure is distinguished by a row of four standing columns made from sea-coral that have capitals carved from the same blocks of stone as the columns themselves (Plates 1&2). Without question, these columns date from before Islam. In addition, they have a markedly Byzantine look to them, and arguably reflect the 6th century Abyssinian occupation of the region18. It is even conceivable that they are still in situ, representing the remnants of a temple that was converted into use as a mu/sallâ in Islamic times. The villagers describe how, from time to time, winds scour away the sands that cover the ground, revealing a paved floor. It speaks of investment in a building feature far beyond the aspirations of the residents of today’s modest settlement. Conceivably the 6th century port-site lies beneath today’s village, just to the east19. The sea has long since migrated far away from the immediate vicinity, and fishermen must cross up to a kilometre of sabkha flats to reach the shoreline.
- 20 Stone, 1985, p. 35; Keall, 1983, p. 389-390.
22To the north, over one kilometre distant from the village of Ghulayfiqa, there is extensive saltpan activity for the production of sea-salt, as in Niebuhr’s day. Here there is a shallow lagoon, and grass flats that likely reflect the scenario of the gradual in-fill of the former harbour. Very close to the flats is the potsherd-covered ground that became the focus of CAMROM’s fieldwork in 2008 (Plate 3). The sherd site had been first identified in 1982, following a tip-off from Francine Stone whose Tihâma expedition had fortuitously camped at the site20. The sherds she produced for CAMROM to observe in Zabîd – classic so-called Hatched Sgraffiato – provided even then the vital clue that she had found a site that likely reflected a medieval settlement site of the famed Ghalâfiqa. Subsequent sherding by CAMROM in both 1982, and as confirmed during casual later visits, with the recovery of identifiable Chinese and Egyptian imports, further revealed that the sherd site could be placed in the relatively narrow time-frame of 9th - late 11th centuries. In fact, in 1982, this reality was pivotal in terms of the ability of CAMROM to develop its ceramic typology for the entire study area. For the imported materials, in their limited time frame of production, allowed for the attribution of similar dates to the locally produced pottery found at the site of Ghalâfiqa. It is important to note that as a result of the 2008 work it can be stated that the fabrics present in these wares found at Ghalâfiqa are slightly different than those recorded at the kiln-site in Zabîd. There is a greater incidence of reduction firing in the sherds found at Ghalâfiqa, indicating production in the style of Zabîd, but using an inferior firing process. The same principle was observed in the 9th-10th century pottery recorded at the Kitf al-A/hmar site of al-Fâzza bay, as described above.
23The sherds on the ground identified by CAMROM are easily visible in open patches between small sand dunes, and spread across ground measuring roughly two hundred metres across (see Fig. 3), though as with all archaeological sites one must be cautious about assuming that this reflects nothing but actual settlement. The site is limited by grass-flats at sea level to the west, and the ground to the east rises uphill. So the extent to which the pottery has migrated away from the settlement site is limited, though the dumping of trash outside the settlement may certainly exaggerate what were once its built confines. But at least we may conclude that we are looking at a settlement covering four hectares at least.
24The archaeological site is dissected by a shallow run-off gulley. Natural stands of dûm-palms attest to a high water table, as do the grass flatlands to the west. Influent streams flow just below ground until they percolate out at the sea’s edge where they sustain the grass-flats. Elsewhere on the site low coppice dunes anchored by sea-thistle vegetation are scattered across many areas of the former settlement. In the central part of the site this amounts to more than half the area. Interspersed between the dunes are exposed surfaces liberally covered with sherds. No discernible difference has been observed in the range of pottery types recovered in any part of the site, either from the central area or from the peripheral ones. Nor have any differences been observed in pottery excavated in the probes that were conducted in 2008. Significantly, and not surprisingly, the probes that were made cutting back into the dunes (Plate 4) revealed much more pristine contextual settings than existed in the exposed areas. Although this pristine context was nothing more than an ashy midden, it speaks nonetheless of contemporaneity for the materials found in it. Here, as in the open, there was no observable difference in the variety of pottery types found.
25The limited range of pottery presented here in this article (Figs.4-6) is chosen to represent classic Zabîd area types, of which many examples were found, as well as the distinctive imports that are highly significant though small in number, as they were at al-Fâzza too. No attempt here is made to create a precise chronology for the site based on the pottery except to observe that Sâmarrâ’ horizon wares are present (though they are largely absent at Sharma), which probably implies a pre-980 date for the beginning of the identified Ghalâfiqa site.
- 21 Cf. Ciuk & Keall 1996, pl. 32d.
- 22 Readers of this article should be aware that many of the dates suggested for pottery illustrated i (...)
- 23 So-called “Trackware”, as seen in Ciuk & Keall 1996, pls. 15, 37, 41-42.
26Amply present also at Ghalâfiqa are jars with 3-prong combing used to effect a simply wavy-line decoration21, rather than the complex patterning of later Trackwares.Because of its occurrence after the single groove wavy-line of the Zabîd kiln site wares, but beforethe later more complex combed patterns, this simple 3-prong combing has been dubbed by CAMROM as “Transitional”, with an 11th century date22. But noticeably absent are the Zabîd Project 3-pronged combed Trackwares23, which are present at Sharma. So we may assume that this particular settlement at Ghalâfiqa ceased to be inhabited around 1100.
- 24 Ibn al-Mujâwir,p. 240.
- 25 Stone, 1985, p. 35 & n. 16.
27This leaves as elusive the second most famous Ghalâfiqa reference, namely the citation by Ibn al-Mujâwir of activities at the port by Persians arriving from Jidda who built a mosque with a minaret at the end of the 11th century. Later, wooden panels taken from the minaret after it had fallen into decay were used for reconstruction work in the al-Ashâ‘ir mosque in Zabîd24. Francine Stone refers to Geniza documents in which there are references to the port of Ghalâfiqa in the 1130s and 1140s25. But no material recovered by CAMROM can be attributed to the 12th century. It is obvious that if these historical reports are true, the settlement lies somewhere other than what has been observed so far.
28If this 12th – 14th century site is to be found, the telltale signs are likely to be fragments of fired bricks and mortar on the surface of the ground. CAMROM has found that any substantial site in its study area dating from the 12th – 14th centuries was built using fired bricks (as with the Rasûlid villa in the al-Fâzza bay). Of course, this hypothetical Ghalâfiqa site is also likely to have been mined for its bricks. All of the 19th century Ottoman forts along this part of the coast were built using recycled bricks. Nevertheless, fragmentary traces will be blatantly visible. Conceivably it will be found when the pattern of sand dune cover changes, as happened with the covering and subsequent exposure of the site of al-Buq‘a in the al-Fâzza bay. So, identification of this settlement awaits discovery.
Fig. 1, Position of Ghalâfiqa on the Red Sea Tihâma coast.
Fig. 2, Shoreline changes and archaeological sites in the Bay of al-Fâzza.
Fig. 3, Sherd scatters at the port settlement of 9th-11th century Ghalâfiqa.
Fig. 4, Pottery from Ghalâfiqa in the style of Zabîd's 9th-10th century kiln production.
Fig. 5, Sâmarrâ’ horizon imported pottery, nos. 1-5; Zabid kiln types, nos. 6-9; painted glazed ware (unique piece), no. 10.
Fig. 6, Red bodied imported glaze wares, nos. 1-6; Chinese imports, nos. 7-9; Egyptian import, no.10; southern Chinese import, no.11.
Pl. 1, Mi/hrâb area of the mu/sallâ outside the village of Ghulayfiqa.
Pl. 2, Detail of the pre-Islamic columns of the Ghulayfiqa mu/sallâ.
Pl. 3, The old harbour (al-sharma) of Ghalâfiqa.
Pl. 4, The sherd scatter area of the 9th-11th century settlement site of Ghalâfiqa.