Summary report

The aims of the Islamic fortifications project on Soqotra were to undertake a comprehensive and multi-disciplinary study of the Islamic fortifications at Jebel Hawari and Firigi and train a team of local archaeologists from the Soqotra Heritage Project (SHP). A team of women and men who have a long-term interest in and knowledge of Soqotra’s rich archaeological heritage. The aims of this project were successfully realised in 2022 between the months of March and April – Jebel Hawari and, due to the political situation in Yemen, January 2024 - Firigi. During these two seasons local archaeologists were trained and became proficient in basic and advanced survey techniques and excavation practices. This included training in the setting up, taking of points and the processing of data from a total station, single context excavation strategies, and artefact drawings and planning. During both seasons the total station survey was coupled with digital aerial imagery taken with a drone to generate a georeferenced orthomosaic base map and create a comprehensive site plan and photogrammetric 3D model of both Islamic forts at Jebel Hawari and Firigi. During the first season at Jebel Hawari the survey and excavation undertaken established that the fort was constructed in four phases, Islamic 1 – 4. The first phase (Islamic 1) was the construction of a rectangular fort, a cistern, and an area of walling between the cistern and fort. This was followed by three successive phases of construction (Islamic 2 – 4). During Islamic 2, the rectangular fort was reinforced along its southwestern edge further walling was added. During Islamic 3, significant defensive structures were added, notably a curtain wall and at least three bastions. The final phase of construction, Islamic 4, was the construction of a second cistern or water catchment feature to the east of the first cistern, the reinforcement of the walls and the addition of enclosures that were presumably animal pens and / or storage areas. During the excavation of a trench slot within the cistern a construction, destruction, and abandonment phase were identified. The destruction phase appears to be related to a burning event from which a charcoal sample was dated to cal AD 1443-1617. Based on the results of the C14 analysis it is likely that the destruction phase of the fort was related to the historical battle between the Portuguese and the Al Mahri in 1507. During the second season at Firigi the survey and clearance established the presence of an Islamic fort that had been mentioned by Bent during his visit in 1899, and briefly sketched by bent Shinnie in 1956. Based on the results of the survey the actual area of the fort is larger than that sketched by Shinnie and is composed of both a revetment wall and four interior enclosures. Within both forts local Soqotri pottery was collected, photographed, and draw for future studies. Overall, the Islamic Fortifications of Soqotra has successfully shed light on Soqotra’s little-known, yet rich, Islamic archaeological heritage and provided local archaeologists with the necessary experience and training to continue this important work in the future.

Detailed report

The Islamic Fortifications Project on Soqotra formed an integral part in the inclusion of Islamic cultural heritage within the World Monuments Fund Watch (WMF) List on Soqotra A program that has sought to raise awareness of the threats facing Soqotra’s cultural heritage and outline potential conservation actions required. Working in collaboration with the General Organisation of Antiquities and Museums (GOAM) Yemen and the WMF this project has successfully identified the risks facing Soqotra’s little-known, yet rich, Islamic archaeological heritage, notably the few remaining defensive structures and ‘Alha Mosque.

The overall aims of the Islamic Fortifications Project on Soqotra have been successfully completed with the second season of fieldwork that undertook a comprehensive and multi-disciplinary study of the remains of the fort identified at Firigi. A fort that was first mentioned by Bent in 1899 and was later sketched by Shinnie, who believed it has been built by the Mahri in the 16th century CE after the departure of the Portuguese (Shinnie 1960: 107, 108, Fig. 6). The fort at Firigi was later briefly visited by Naumkin (1993) and Weeks et al. (2002), both of whom confirming Shinnie’s findings that, due to its location, it functioned as an important base from which the Mahri could move further into the island or make punitive raids against the indigenous Soqotri inhabitants. Working alongside the director of GOAM and a local GOAM representative together with team of local archaeologists from the Soqotra Heritage Project (SHP) and experts in pottery and geometric surveys, the second season undertook a comprehensive survey of the fort at Firigi. To facilitate the training of local archaeologists’ experts were assigned to each of the SHP archaeologists to assist in the development of their skills in undertaking geometric surveys and analysing artefacts, notably the pottery from the site.

The first phase of the second season was generate a sequence of base maps using digital terrestrial photography coupled with a GPS enabled drone and a total station to georeference the base maps. Three successive base maps were generated, 1. before the clearance of vegetation and areas of overburden - where a limited number of features were observable, 2. after clearance of vegetation and areas of overburden - where features were better defined, and 3. after further clearance of collapse and overburden – where an architectural sequence for the fort could be clearly defined. The collating and georeferencing of the various images allowed for a sequence of georeferenced orthomosaic base maps to be generated. These base maps were then incorporated within a singular model from which stratigraphic units could be identified and a typological, technical, and architectural sequence for the fort established. This allowed for the identification of a preliminary architectural sequence that can be correlated with the construction techniques and architectural style of the fort. In addition to the creation of a georeferenced 3D model of the fort, a georeferenced 2D base map was created within a spatial database, QGIS (Figure 1).


Figure 1. A georeferenced orthomosaic of the Fort at Firgi.

The aim of the second phase of this season was to establish a chronology and locate evidence that would allow for the establishment and abandonment phases of the fort to be determined. Large parts of the site had been disturbed by an adjacent date palm garden that has resulted in loss of structural elements from much of the western part of the site, while along the northern extent large parts of the site have been subject to erosion from the planting of palm trees at the base of the outer wall. During initial surveys and clearance of the site it was clear that the structures of the eastern area of the fort had been built on a platform that raised the eastern area and provided a level platform for the construction of the walls, towers and other structures that were recorded. The platform much like the structures within the fort were constructed from stones from the nearby wadi bed. While it was evident that a soil layer had formed on some parts of the platform, notably where the bedrock was incorporated as part of the base for the platform, much of the sediments within the fort have been washed out leaving behind little dateable material. Nevertheless, during clearance of the overburden and collapse from the northernmost tower structure several fragments of local Soqotri pottery were collected. Dating these local wares is particularly difficult in that they have been in continual use throughout the last two millennium and virtually no studies have been undertaken to establish either a chronological or typological sequence. Nevertheless, establishing a local ware typology from the pottery collected at Firigi and that at Jebel Hawari is an important step towards better understanding local wares and potentially establishing relative chronology. As such the pottery collected is currently being analysed by a pottery expert that came to Soqotra for the second season.

The third phase of the second season sought to collate and digitise the written and photographic site records of past survey and excavation works. These were then incorporated within the registers, plans, sections, matrices generated during this season’s fieldwork. The digitisation of this site archive is currently being completed and incorporated within an open-source spatial database (pyArchInit for QGIS) for further analysis. While the results of this season’s work are preliminary and further analysis is still necessary, several key points concerning the constructional and architectural phases of the fort have been established.

During our initial large-scale landscape survey we were able to ascertain that the fort formed part of a larger structural complex, which was built upon a natural wadi bank that formed along the western edge of the wadi. At present it is not clear as to when this wadi bank formed, but it was evident from the various structures that some form of agricultural was being practiced. According to interviews with the local inhabitants these structures were ‘ancient’, which typically means they existed over two generations ago. While many of these structures have been reused by the local inhabitants, either as walls to prevent the goats from accessing the date gardens within, or as goat pens, it is still possible to observe the presence of fields that are outlined by boulders. The boulders surrounding the fields appear to have been used to contain the soil, which is much deeper within these areas, presumably for cultivating cereal crops. At present these fields are predominately used by the local inhabitants for date palm cultivation, although in several areas they are also growing fruit trees, vegetable plots, and finger millet. While it is not certain as to what crops may have been used in the past, with the exception of the date palms that have been recorded from at least the thirteenth century (Ibn al-Mujāwir 1954: 2.265), it is likely that some form of agriculture was being practiced by the inhabitants in the area. How these cultivated fields relate to the fort is not clear, yet one can surmise that they would have potentially been used to supply the garrison stationed at the fort with much needed supplies. Indeed, if the Mahri garrison were engaged in controlling trade and conducting punitive raids it would be unlikely that they would have been supplied by the local inhabitants. As such, they would have needed to cultivate food for themselves, although one cannot of course rule out that this may have also provided them with a resource that they could have bartered with. That the fort lies to the south eastern edge of these fields it is unclear how they would have guarded the fields against incursions from the local Soqotri inhabitants, if this was required. While no direct evidence for housing or similar structures was located in or around the fields, it is possible that these structures could have either been robbed, or converted into goat pens.

The second stage of our survey was to compare Shinnie’s plan of the fort with the georeferenced orthomosaic and observations made during this seasons work at the fort. According to Shinnie (1960: 107-108, Fig. 6), along the front wall of the fort a ‘type of moat’ had been constructed by artificially enlarging and deepening the wadi. However, the wadi bed in front of the front wall of the fort and several meters away from the wall ends directly on the bedrock, which bears no marks of having been enlarged or deepened. That a pool may have formed in front of the fort is more likely related to the nature of the wadi bed that at the point where the embankment wall flows past the fort is deeper, due to the bedrock formation rather than any human agency. Another aspect of the fort relates to its construction, notably the use of ‘trimmed blocks, quite unlike the local building style’ that Shinnie (1960), Naumkin (1993) and Weeks et al. (2002) highlight as being further evidence for the site having been constructed by the Mahri. Closer inspection of the red granite stones used in the construction of the fort provide a bit of a conundrum in that the majority of the construction of the fort is made up of both rounded and sub-angular cobbles of red granite. Within the fort the flat surfaces of the sub-angular and angular cobbles that were used in the towers to create a flattish face, while elsewhere a mixture of rounded and sub-angular and angular cobbles were used in the forts construction. Based on the surrounding geology the rounded cobbles would have been procured from the wadi, while the sub-angular and angular cobbles would have been procured from the sides of the mountains to the north and south respectively. The fact that sub-angular and angular rocks were used in the construction of the facing of the towers of the fort is not a testament to either a Mahri or Soqotri construction but rather a testimony to the innovative use of local materials for the construction of the fort.  

Based on Shinnie’s plan the fort is triangular in shape and made up of two towers, in the north and south respectively, a well in the centre of the fort and a series of adjoining rooms that are related to the northern tower (Shinnie 1960: 108, Fig. 6). However, once the vegetation and overburden had been cleared it was clear that the fort was actually rectangular in shape. Moreover, the so-called well was, according to the local inhabitants, not a well but was used as a goat pen to prevent young goats from straying. Once the vegetation had been cleared within the circle of rocks it was clear that the stones had been built on top of the ground level of the fort and that the so-called well did not go any deeper than the floor of the fort. During the documentation of the eastern most wall, however, a stream of water coming from under the fort was noticed. This stream was said to come from a spring that was situated further west of the fort, although it was not located. It is highly likely that this may have been a source of water for the inhabitants of the fort in the past.

In the northern part of the site the remains of a square double-walled tower were identified. Based on initial observations the outer wall was built later and could have functioned as a support for the inner tower. After clearance of the collapse from the southern wall the tower we have ascertained that the tower would have been at least two meters in height from the base of the fort. While it was not possible to remove the collapse from the interior of the tower it was clear that the interior been accessed from the west, where it formed part of a rectangular structure. This large rectangular structure is made up of four square rooms, all of which have collapsed leaving little more than the foundational stones in place. The rooms are all approximately 2.5 by 3.2 meters in size and appear to have either functioned as storage or living quarters. Only small fragments of local Soqotri pottery were found within the overburden.

The southern part of the site consists of the remains of a tower, much of which has been damaged by the planting of a date palm tree. To the north of the tower is a rectangular room measuring 5.2 by 5 meters of which only the foundation remains. No finds were made within the room. The tower and rectangular room are raised above the floor of the fort by approximately 30 cm and, based on the subsidence in the north-western corner it appears as if this had been built up with earth, which may have been used to create a floor on the bedrock, which provides this area of the fort with a natural elevation.

As mentioned previously, the western part of the fort has been partly destroyed during the planting of a date palm plantation, and the robbing of stone to construct a wall to the garden to prevent goats from entering. However, during a short investigation into this heavily overgrown part of the site structures related to the western edge of the fort were found. These were the remains of foundations, presumably for rooms and, at the far western edge of the garden, walls.

In summary, while the analysis of the results from this season’s work remain preliminary it is clear that we have located the Al Mahri fort that Shinnie (1960) identified. Moreover, based on the initial overview of the fort we were able to ascertain that, contrary to Shinnie’s site map, the fort was rectangular in shape with two towers in the north and south-eastern corners. Due to the underlying bedrock on which the fort was built, it is clear that the central and part of the north-western edge of the fort had been built up and levelled with boulders from the wadi. Due to the garden in the western edge of the fort it is not clear as to the extent of the fort, nor whether it was levelled. Of particular interest was that the construction of the fort. While historically and ethnographically the fort at Firgi is believed to have been occupied by the Mahri, that the fort was built specifically by the Mahri is unclear. Contrary to the reports by Shinnie (1960), Naumkin (1993) and Weeks et al. (2002) the fort was not constructed of ‘trimmed blocks, quite unlike the local building style’, but was actually constructed from unworked red granite from the wadi and surrounding mountain slopes. This could imply that the Mahri used local labour in the construction of the fort, or that they simply made use of the most easily accessible material available.