Tantura F is the first wreck excavated so far in the eastern Mediterranean dated to the beginning of the 8th century AD (the local early Islamic period), based on 14C and pottery analysis. Among the finds were the remains of about 30 ceramic vessels, two anchors, food remnants, fish-bones, a needle, a spoon and a glass vessel. Tantura F was c.15 m long and 5 m wide, and hull remains comprise keel, frames, planks, stringers, mast-step, and other internal components. Construction features clearly indicate frame-based construction, considerably earlier than it is generally thought to have been first employed.
Tantura F was discovered in 1996 during a survey at Dor (Tantura) lagoon (on the Mediterranean coast of Israel, about 30 km south of Haifa) by a combined expedition of the Institute of Nautical Archaeology at Texas A&M University (INA) and the Recanati Institute for Maritime Studies at the University of Haifa (RIMS), headed by S. Wachsmann (Wachsmann et al., 1997: 7). The wreck was designated ‘Trench 10’ or ‘Tantura 10’, and recently ‘Tantura F’. In the 2004 season it was excavated by a combined expedition of the Nautical Archaeology Society (NAS) of Great Britain headed by C. Brandon, and the RIMS headed by Y. Kahanov, together with K. Raveh.
The wreck is orientated roughly north-west/south-east, about 70 m offshore, a few metres north-west of the lagoon's navigable channel. It is in 1 m of water, buried under an additional 1.5 m of sand mixed with shells and stones. Although some of the stones lay directly on the ship's timbers, it is not likely that they were part of the cargo or ballast. The archaeological remains spread over an area of 12 × 3.5 m (Fig. 1). Based on 14C tests, some by the AMS method on short-lived organic materials,1 and ceramic analysis, the wreck was dated to the beginning of the 8th century AD, which is the local early Islamic Umayyad period. Excavation is still in progress, and some timbers remain to be exposed, measured and recorded.
Wood samples were taken from every hull component for species identification by N. Liphschitz of Tel Aviv University. The hull was made of three different species: Tamarix X5 (frames and central longitudinal timbers), Pinus brutia (keel section, frames, planks, stringers, and mast-step assemblage), and only one example of Pinus nigra (keel section). The Tamarix X5 originated in three areas of Turkey: western (Izmir area), north-west (Edirne area), and southern Turkey (Antalya area); and the Pinus brutia originated in western (Izmir area) and southern Turkey (Antalya area). A mat that was used for protecting ceramic vessels was made of Phragmites communis, which also grows in Turkey.
The hull survived up to the turn of the bilge, and almost to the bow and stern. Its remains comprised the complete keel, frames (floor timbers, futtocks, and pairs of half-frames), sections of strakes from port and starboard sides, stringers, mast-step assemblage, central longitudinal timbers, and several other internal components (Fig. 2).
The keel was made of Pinus brutia and Pinus nigra. Where measurable, its average dimensions were 95 mm sided and 160 mm moulded (Fig. 3). No rabbet or chamfered corners for fitting the garboards were evident, but rabbets were identified at its ends. Remains of scarfs were found at the bow and stern ends of the keel. It is still unclear exactly which components were scarfed: the keel and the stem, the stem with its extension, or something else (Fig. 4). An additional scarf was found in the keel between frames F7 and F8, but its details are still hidden. The use of at least two scarfs and two different tree-species in the keel may perhaps hint that it was difficult to obtain long timbers suitable for the keel.
There was evidence of 36 frame-stations, which was apparently the total number. The timbers of five frames did not survive, and they were identified from staining patterns on planking surfaces and nail-holes. The 31 extant frame-timbers were exposed, including floor timbers, pairs of half-frames, and futtocks. They were made of Tamarix X5 and Pinus brutia. Generally, the framing pattern was of alternating floor timbers and half-frames, except under the mast-step, where a series of floor timbers and futtocks only was identified. Amidships, frames and floor timbers were flat; their angle of deadrise increasing towards the bow and stern. The angle of deadrise of the frame at the bow was 43°. The frames, of variable length, were on average 8 cm sided by 11 cm moulded, with room and space of 28 cm. They were fixed by iron nails to the keel.
Sections of 15 futtocks were recovered, fastened by iron nails to the floor timbers from the sides. They were fitted randomly forward or aft of their associated floor timbers (Fig. 5). Half-frames were scarfed to each other by side joinery, and were further connected by iron nails. Limber-holes were found 20 to 100 mm from each side of the keel and at the turn of the bilge. They were rectangular or semi-elliptical, with average dimensions of 30 × 35 mm or 30 × 20 mm respectively. Some of the limber-holes at the turn of the bilge penetrated the floor timber or half-frame and futtock. Rope fragments were found in one of them. Upper surfaces of frames were covered with matting, which apparently served for amphora storage and for the protection of the frames—there was no evidence of ceiling planking.
Planks were made of Pinus brutia, and were connected to the frames by square iron nails of 5 mm average side, driven from the outside (Fig. 6). One or two nails connected each plank to each frame, depending on plank width. Planks were 8 to 20 cm wide, and their average thickness was 25 mm. No planking edge-joints were found anywhere, although not all the edges of the planks have been exposed at this early stage of the project. Caulking was found in many places between seams. Black staining, apparently the result of some kind of fire treatment, and several resin layers, were identified on the inside and outside surfaces of the planks. The garboards butted the upper sides of the keel, but were not connected to it. The garboards fitted into rabbets only at the ends of the keel. So far only one plank butt scarf under a floor timber has been identified.
Six stringers made of Pinus brutia half-logs were found (Fig. 7). Two central stringers were connected to the upper surfaces of the frames, with their flat surfaces upward, located on either side of the longitudinal axis of the ship, about 5 cm apart. The mast-step was installed between these stringers. The other four stringers, two on either side, were fixed with iron nails to the frames with their flat surfaces downward, fitting the upper surfaces of the frames. Average dimensions of the stringers were 15 cm wide by 6 cm thick.
Central longitudinal timbers
Two central longitudinal timbers made of Tamarix X5 were found at the bow and the stern (Fig. 8). They were fastened with iron nails to some of the frames from above, and notched on their undersides for fitting onto the frames. The timber at the bow, which stretched from frame 32 to frame 27, was 140 mm wide, 120 mm thick and 2.06 m long. The stern timber stretched from frame 3 to frame 7, and was 140 mm wide, 155 mm thick and 1.42 m long. These timbers were worked on all sides.
The mast-step assemblage comprised the mast-step itself and two lateral sisters (Fig. 9). All components were made of Pinus brutia. The mast-step was not nailed to the stringers beneath it, or to the hull. Measurable dimensions were 1.45 m long, 260 mm wide and 200 mm thick. There were two mortises in its upper surface. The smaller forward one measured 100 × 45 mm and 55 mm deep, and probably secured a stanchion supporting a mast-partner. The after mortise, measuring 230 × 50 mm, sloped 22 mm downwards towards the south-east. This mortise probably served to secure the mast heel; and its slope may hint at the direction of the bow and stern, the bow being towards the north-west.
Saw marks were evident on the frames, planks, central longitudinal timbers, stringers, and mast-step. Plank surfaces were later planed. Other saws, apparently handsaws, were used for fitting shell components, where chisel-marks were also discernable. Drilled holes were evident on the central longitudinal timbers.
The finds included pottery; metal objects—nails and two anchors; organic materials—rope, reed matting, a wooden spoon, a bone needle (Fig. 11), a wooden ring, a large amount of fish-bones, and food-remnants (carobs, and pits of Nabali olives, which were common in the central mountains of Israel and mostly known as pickled olives) (Fig. 10). These were analysed by M. Kislev of Bar Ilan University. The pottery assemblage contained about 30 vessels: eight amphoras, some of them similar to the type found in the 7th century shipwreck at Yassi Ada (Van Doorninck, 1989: fig. 1: 11) and Caesarea, Stratum 8, which is dated to the local early Islamic period (Arnon, 2003: pl. 74.11) (Fig. 12); two juglets similar to items found in Kellia in the Nile Delta (Borel and Cattin, 1999: fig. 491: 236, 247), which were produced from the end of the 7th century, after the Arabic conquest (Fig. 13); and 20 storage jars (Fig. 14). The latter were of a type recorded in Caesarea, Stratum 8 (Arnon, 2003: pl. 74: 10), along the Israeli coast (Zemer, 1977: pl. 22: 61–2), Pella, Jordan, from an early 8th-century deposit (Smith, 1973, pl. 32: 489), Amman Citadel, Jordan (Sauer, 1986, fig. 2: 48), and Kellia in the Nile Delta (Types 187–90) from the 8th century (Egloff, 1977: pl. 22: 3). The jars from the coast of Israel and Pella were almost certainly manufactured in the lower Nile region (Watson, 1995: 319). The jars had resinous linings and contained a residue of small fish-bones, which could have been a fish product. The fish remains were analysed by I. Zoar, of the Recanati Institute for Maritime Studies at the University of Haifa, and identified as Tilapia (St. Peter's Fish).
The outstanding state of preservation of the ship's hull, both in the amount of extant wood, and its high quality, as well as the large quantity of finds, makes this one of the most promising sources of data gathered from any ancient shipwreck. It provides essential information about ship construction, and sheds light on Dor and its region of the period. There is no historical or archaeological evidence regarding Dor (Tantura) and the conditions of the society and economy in this region in the early Islamic period. The historical references to Dor from the 7th century AD until the Crusades in the 12th century add no knowledge about Dor and its region, since they are based on previous sources. Not a single reference by classical Arab geographers from the 9th to 12th centuries mentions Dor or Tantura (Dahl, 1915: 108–21).
The abundance of finds suggests that this may have been a fishing vessel. The needle could have served for repairing fishing nets. The large amount of fish remains in the hull further support this idea. However, the existence of at least 20 storage jars with a residue of small fish-bones, the two juglets similar to Egyptian types, the amphoras used in maritime trade, carobs, and the Nabali olive pits, may be evidence of maritime connections, and, more specifically, a sailing route between Egypt and Dor in the 8th century AD. A vessel would not enter Dor/Tantura lagoon unless it had detailed local knowledge and a specific purpose or maybe in an emergency. This may indicate that Dor was inhabited, and this is the first evidence of the existence of a community at Dor at this time. Further analysis of the finds will contribute to knowledge of the economic activity of the period and the connection of Dor with nearby areas.
The Tantura F hull was based on a keel, frames nailed to the keel, thin planks nailed to frames and later caulked, but with garboards not connected to the keel, and several longitudinal reinforcements. There was no trace of shell-first or planking-orientated construction in the hull; specifically no plank edge-joints were found. Thus, this hull is clearly of frame-based construction, and not plank-based. Until recently it was accepted that the Serçe Limanı wreck found in southern Turkey, and dated to AD 1025, was the earliest ship built according to the criteria of frame-based construction. The Serçe Limanı was 15.66 m long with a 5.2 m beam. All frames crossed the keel and were fastened to it with iron nails. The garboards were not connected to the keel except in the ends. Planking was nailed to the frames without planking edge-joints and treenailed in repairs. The hull was longitudinally reinforced by three stringers and two wales, made of half-logs, and a large keelson that was bolted to the keel with forelock bolts. Although less than 3 m of the keelson survived it has been assumed that it extended to the ends of the ship because of discoloration and pressure-marks on frame-tops and bolts running through the keel from the bow to the stern. Pitch was found on the inner hull surface, apparently used as caulking (Steffy, 1982: 13–34; Steffy, 1994: 85–91; Matthews and Steffy, 2004: 81–122; Steffy, 2004: 153–69). Analyses of other ships excavated at Dor lagoon: Dor 2001/1, Tantura A, and Tantura B, show that this method of building already existed in the second half of the 1st millennium AD, and suggests an earlier date for the transition in ship construction from shell-first to frame-based.
Comparison with other Dor lagoon wrecks
Dor 2001/1 is dated to between the 5th and 6th centuries AD. The original length of the ship is estimated at 16 m, and its beam 5 m. The hull remains included the keel, false keel, frames, planks, wales, ceiling planking, stringers, part of the mast-step assemblage, and a central longitudinal timber. The ship was built based on frames. Frames were joined directly to the keel by rectangular metal nails, and the planks were attached to the frames by 6–8 mm rectangular-section tapered iron nails. The garboards were not connected to the keel. Planks were butt-jointed into strakes at frame-stations; and planking seams were caulked. There was no evidence of mortise-and-tenon, or any other variant of planking edge-joints. The keel, false keel, central longitudinal timber, central stringers, foot-wales, chine strake and two wales provided the longitudinal integrity of the hull (Mor, 2002–3; Mor, 2004; Mor, 2005).
Tantura A is dated to between the 5th and the 6th century AD. The original length of the ship is estimated at 12 m, and its beam at 4 m. The hull remains included the keel, an endpost, frames, and planking. The general construction was built on frames nailed to the keel. The garboards were not connected to the keel, but were fastened, as were other planks, to the frames with iron nails. Planks were butt-jointed into strakes at frame-stations, and planking seams were caulked. There were no plank edge-joints of any kind (Kahanov and Breitstein, 1995a; Kahanov and Breitstein, 1995b; Kahanov and Royal, 1996: 21–3; Wachsmann and Kahanov, 1997: 3–15; Kahanov, 2001: 265–71; Kahanov et al., 2004: 118–27).
Tantura B is dated to the 9th century AD. The length of the original ship is estimated at about 18–23 m, and its beam at 5 m. The hull remains included the keel, frames, planks, keelson, mast-step, and stringers. Each frame was fastened to the keel with a single nail, and further reinforced by the keelson from above. Rectangular iron nails 5 × 5 mm were used to connect planks to frames. Planks were joined into strakes at frame-stations with butt scarfs and L-shaped joints. Caulking was found in planking seams. No planking edge-joints were discovered anywhere (Wachsmann et al., 1997; Kahanov, 2000; Kahanov et al., 2004). Table 1 summarizes construction features relating to the building methods of the four ships found at Dor lagoon from the 5th to the beginning of the 9th centuries AD.
Table 1. Summary of hull construction details
End of 5th C. beginning of 6th C.
End of 5th C. beginning of 6th C.
Beginning of 8th C.
Beginning of 9th C.
Based on frames
Based on frames
Based on frames
Based on frames
Keel (sided × moulded, mm)
110 × 160
110 × 180
95 × 160
95 × 104
No keel rabbet, rabbeted endpost
No keel rabbet, rabbeted endpost
Rabbet only at the ends
Keel rabbet only at the ends
Keelson (sided × moulded, mm)
Central longitudinal timber, 150 × 180
Two central longitudinal timbers Bow: 140 × 120, 2.06 m long, Stern: 140 × 155, 1.42 m long
122–202 × 157–180
Stringers (width × thickness, mm)
Two, 125 × 85, 125 × 80
Six, 150 × 60
Two, 70 × 90
Wales (width × thickness, mm)
Two, 210 × 105, 180 × 70
One, 100 × 85
Frames (sided × moulded, mm)
80–85 × 96–120
90 × 95
80 × 110
90 × 94
Planking thickness (mm)
No mortise or tenon
No mortise or tenon
No mortise or tenon
No mortise or tenon
The Dor 2001/1, Tantura A, Tantura F and Tantura B hulls were designed according to the criteria of frame-based construction. They were based on frames nailed to the keel, planking nailed to the frames and butt-jointed into strakes at frame-stations with caulking in the seams, and there is no evidence of mortise or tenon, or any other planking edge-joints. This analysis shows that this method of building already existed in the second half of the 1st millennium AD, and suggests an earlier date for the transition from shell-first to frame-based ship construction than is generally accepted.
The Tantura F wreck is the first early-8th-century vessel discovered and excavated so far in the eastern Mediterranean. The wreck was dated by AMS radiocarbon analysis. The ceramic assemblage of the local early Islamic period from the southern Levant supports this dating. The preservation of Tantura F and her cargo, which included storage jars with fish-bones and food remains, demonstrates that Dor lagoon served as an anchorage for coastal commerce at that time. Whether or not she was a fishing vessel, this sheds more light on the local community, apparently involved in these maritime activities. This is significant, as background information on Dor (or Tantura) of the period is poor, and historical and archaeological sources are absent. The Tantura F hull is clearly of frame-based construction. It is new and essential archaeological evidence for the early completion of the transition in ship construction from shell-based to frame-based in the Mediterranean.
AMS radiocarbon dating was carried out by G. Bonani of the Institute of Particle Physics, Zurich. Sample numbers are (all ETH numbers, uncalibrated): ETH-31266 (1340 ± 45), 31267 (1255 ± 45), 29800 (1360 ± 40), 29801 (1280 ± 40), 29802 (1300 ± 40), 29803 (1340 ± 40), 29804 (1320 ± 40).
This research was supported by Lord Jacobs of London, the Israel Science Foundation, the Hecht Foundation, the Sir Maurice Hatter Fellowship for Maritime Studies, and the University of Haifa. The authors would like to thank Mr. Chris Brandon for his long-term cooperation and the drawings, and to Mr. John Tresman for reviewing the manuscript and the English editing.