The 7th–9th Century Tantura E Shipwreck, Israel: construction and reconstruction



The Tantura E shipwreck was discovered in Tantura lagoon, Israel, in a water depth of 2.8 m, covered by 1 m of sand. It was a coaster that plied the Levant coast, dated to the 7th–9th centuries CE. It was recorded under water, but several components were studied on land, as well as the finds. The hull is of frame-based construction, with flat floor timbers and a sharp turn of the bilge. The archaeological evidence, as well as modern design tools and regulations, suggests that Tantura E was c.12.5 m long, of 25 tonnes displacement, and could load c.17.5 tonnes.

Dor (Tantura) lagoon, where many shipwrecks have been brought to light (Kingsley and Raveh, 1996; Wachsmann and Kahanov, 1997; Wachsmann et al., 1997; Kahanov, 2011), is situated on the coast south of Tel Dor, about 30 km south of Haifa, Israel (Fig. 1). The Tantura E shipwreck was discovered in 1995 during a survey by a combined expedition, headed by S. Wachsmann of the Institute of Nautical Archaeology at Texas A&M University (INA), and the Leon Recanati Institute for Maritime Studies at the University of Haifa (RIMS), and denoted as ‘Trench IX’ (Royal and Kahanov, 2000). The shipwreck was excavated systematically in 2006, 2007 and 2008 by the RIMS, directed by Y. Kahanov, with the participation of volunteers of the Nautical Archaeology Society of Great Britain (NAS) headed by C. Brandon, and K. Raveh of the local Aqua Dora diving centre (Israeli and Kahanov 2011; Israeli and Kahanov, 2012).

Figure 1.

Dor/Tantura lagoon and the Tantura E shipwreck site. (N. Yoselevich)

The shipwreck is located c.25 m offshore on the western side of the lagoon's navigable channel at 32°36′31″ N, 34°54′56″ E. It is oriented north-west/south-east, in a maximum water depth of c.2.8 m, covered by 0.7–1.0 m of sand. The water depth and the thickness of the sand layer change during the seasons, as a result of sand movements resulting from varying sea conditions—wind, waves and currents, typical to the lagoon. Apparently the wooden hull remains had been exposed at times and had deteriorated considerably over the years.

The hull remains are spread over an area of 7.6 × 3.1 m, and the dimensions of the original hull have been estimated at 12.5 m long, with a maximum beam of 4 m. It was recorded under water, and several selected timbers removed and studied in the laboratory at the RIMS.

Among the finds were building stones, bricks, a clay disc, lead fishing weights, and abundant pottery sherds (not all in situ) from a wide span of periods, the majority dated to the 4th–7th centuries CE. Based on ceramic typology and 14C analyses, the wreck has been dated to between the 7th and 9th centuries CE, which spans the local late Byzantine–early Islamic periods.

The significance of the hull of Tantura E is in its construction from many small timbers and its being frame-based, which is early for this construction feature (Pomey et al., 2012).


About three-quarters of the bottom of the wooden hull survived (Figs 2, 3). It was found in a poor state of preservation, made of many small timbers, generally frames and planks, with many repairs. The ship was preserved up to the turn of the bilge on both sides. The north-western end of the shipwreck and the north-eastern side survived in a better state of preservation than the south-eastern end and the south-western side. The surviving timbers comprised sections of the keel, false keel, two transitional timbers (a connecting timber and a gripe), a fragment of one endpost at the north-western end, a small fragment of an extension of the keel at the south-eastern end, a central longitudinal timber, 59 framing timbers and frame fragments, 23 strakes (13 on the south-western side and 10 on the north-eastern), 8 stringers, 11 ceiling planks, a bulkhead support, and a stanchion.

Figure 2.

Tantura E, general plan of the hull remains after the 2008 season. (C. Brandon)

Figure 3.

Tantura E, overview from north-west end to south-east, during 2007 season. (A. Yurman)


A 7.6-m length of the longitudinal backbone of the hull survived, comprising five timbers (K1–K5) (Fig. 4): K1 a short fragment of the endpost at the north-western end; K2 a transitional timber (gripe); K3 another transitional timber (connecting timber); K4 the main component of the longitudinal spine; K5 a short section of another timber at the south-eastern end that could have been an extension of the keel, another transitional timber in the south-eastern end of the hull, or an endpost.

Figure 4.

The keel, its components (K1–K5) and scarfs (S1–S4). (E. Israeli)

These longitudinal backbone components had a rectangular cross-section, 100 mm sided, and 170 mm moulded on average (Table 1). Four of the five components comprising the longitudinal backbone of the ship remains: K1, K2, K3, and K5, were retrieved and studied in the laboratory. Three of these timbers (K1, K2 and K3) were connected successively at the north-western end of the main keel timber (K4); and one (K5) was connected at the south-eastern end.

Table 1. Components of the longitudinal spine. All dimensions in mm
ComponentLengthSidedMouldedTree species*Notes
  1. * Liphschitz, 2006, 2007a, 2007b, 2008, 2010.
K127595154Cupressus sempervirensBroken. Bolt hole
K2124089119115156192170Cupressus sempervirensBolt hole in scarf no. 2
K32090117135120139161160Cupressus sempervirensBolt hole in scarf nos 2 and 3
K44740120150Cupressus sempervirensBolt hole in scarfs nos 3 and 4
K51618055Tamarix sp.Broken

Endpost K1

This is a fragment of the endpost surviving at the north-western end of the spine, (Fig. 5a). It measured a maximum of 275 mm long, 95 mm sided and 154 mm moulded. The endpost was connected at its south-eastern end to the transitional timber (gripe, K2) by a vertical tenon at its end, fitting a mortise in the transitional timber. A bolt hole 14–16 mm in diameter penetrated the gripe (K2) from below through the entire thickness of the endpost perpendicular to its upper surface (Fig. 5b).

Figure 5.

X-ray of the components of the north-western scarf (S1) in the longitudinal backbone. a) KB1; and b) KB2 are remains of the bolt holes. (S. Shalev and E. Israeli)

Transitional timber K2

The second spine timber from the north-west was a transitional timber, a gripe (K2) (Fig. 6). It was 1.24 m long, slightly curved in the vertical aspect, and averaged 115 mm sided and 170 mm moulded. It was connected to the endpost at its north-western end, and to the second transitional timber (connecting timber, K3) at its south-eastern end. Evidence of several bolts and nails was found in the upper and lower surfaces of this timber. Nails connected frames to the gripe from above, and the false keel to the gripe from below. At both ends bolts, driven from below, connected the keel components to each other. The bolt heads were countersunk to facilitate the fixing of the false keel. A single false-keel section was found connected to the underside of the transitional timber (gripe, K2, Fig. 7). The false keel was made of Cupressus sempervirens, 1.14 m long, 110 mm wide and 40–55 mm thick.

Figure 6.

Transitional timber (gripe), K2. (J. J. Gottlieb)

Figure 7.

Transitional timber K2, and the false keel, view from north to south. (A. Yurman)

Transitional timber K3

The third component from the north-west end of the longitudinal backbone of the hull was another transitional timber, K3 (Fig. 8). It was connected at its north-western end to the transitional timber (the gripe, K2), and at its south-eastern end to the main timber of the keel (K4). Both joints were complex keyed hook-scarfs. This timber was slightly curved, 2.09 m long, and averaged 120 mm sided and 160 mm moulded. Evidence of bolts and nails was found in the upper and lower surfaces of this timber. Nail remains in the underside are evidence of the existence of the false keel, which did not survive in this area. Damage was discerned on the underside of the connecting timber. The false keel was probably lost during the ship's operating life, leaving the connecting timber unprotected from below.

Figure 8.

Transitional timber (connecting timber) K3. (R. Gilboa)

Main section of the keel K4

This was a straight timber 4.74 m long, the longest identified in the shipwreck, averaging 120 mm sided and 150 mm moulded. It was studied under water. In three places the underside of the keel was excavated and uncovered completely. Its lower side was found to be worn and eroded, with rounded edges, which probably happened during the operating life of the vessel. No remains of the false keel were identified under the keel timber (K4) in the areas excavated. Its upper surface and sides were in a reasonable state of preservation.

A rabbet was identified in all keel timbers along the entire length of the spine (Fig. 9). The upper edge of the rabbet was on average 32 mm below the keel's upper surface. The rabbet was on average 24 mm, the back rabbet 13 mm, and c.10 mm deep into the surface of the side of the keel. The rabbet corresponds to the plank thickness (see below). Nail remains connecting the garboards to the rabbet were found in K2, K3, and in the south-eastern end of K4. The angle of deadrise varied from about 22° at the gripe to about 6° amidships.

Figure 9.

a) Rabbet in K4 (A. Yurman); b) cross-section (E. Israeli)

Timber K5

At the south-eastern end of the longitudinal spine was a short section of another timber (K5) (Fig. 10). This timber was hook-scarfed to the main keel timber (K4), and fastened by an iron nail, of which only a large concretion remained. It was made of Tamarix spp., unlike all the other spine components, which were of Cupressus sempervirens. This timber was maximum 161 mm long, 80 mm sided and 55 mm moulded. This fragment was too short to determine whether it was a straight continuation of the keel, a transitional timber, or an endpost. If symmetry is assumed in the longitudinal spine (although no archaeological evidence for this exists), the timber would have been a section of a transitional timber.

Figure 10.

K5—south-east remains of the longitudinal spine, scarf S4 and suggested reconstruction (dotted line). (A. Yurman and E. Israeli)


Four scarfs (S1–S4) (Fig. 4) were identified and documented along the longitudinal backbone.

Scarf S1 (Figs 4-7), near the north-western end, connected the endpost (K1) with the transitional timber, the gripe (K2), by a vertical tenon at the end of the timber. The tenon was 150 mm wide. Its thickness varied: at its base it was 27–29 mm, and at its external end 25–27 mm. It was slightly tapered both in thickness and width, and its length varied between 30 and 40 mm. The mortise cut in K2 fitted the tenon, and on both sides were cheeks 38 mm thick. No evidence of a nail connecting the scarf members to each other was found in this scarf; however, bolt remains on each component were evident outside the scarf. These bolts are evidence of a reinforcing timber—a central longitudinal timber (CLT) or its continuation—that held the scarf components together.

Scarf S2 (Figs 4, 6, 8, 11), 1.16 m from the north-western end, connected the two transitional timbers (K2 and K3). It was a keyed hook-scarf, 300 mm long, which was found with its two components firmly attached. The key was rectangular, 40 mm wide and 14 mm thick. Remains of an iron bolt of square cross-section, calculated at 11.2 × 11.2 mm, penetrated the wooden key of the scarf. This scarf was reconstructed at full scale in the wood workshop at the University of Haifa (Fig. 11).

Figure 11.

a) S2, the scarf in the north-western component—K2; b) reconstruction, turned 90° counter clockwise. (E. Israeli)

Scarf S3 (Figs 4, 8), 3.08 m from the north-western end, connected the transitional timber (connecting timber, K3), with the main section of the keel (K4). It was a keyed hook-scarf, 310 mm long, with its two sides still attached. An iron bolt hole, 16 mm in diameter, penetrated the wooden key of the scarf. This scarf was similar to S2.

Scarf S4 (Figs 4, 10), at the south-eastern end, connected the main keel timber (K4) with K5. It was a keyed hook-scarf, c.400 mm long, in a bad state of preservation. A large iron concretion indicates that this scarf was also fastened by an iron bolt.

Central longitudinal timber (CLT)

A CLT made of Pinus brutia was found in the north-western area of the shipwreck, above the frames. Its remains extended between scarfs S2 and S3 (Fig. 12). It was retrieved and studied in the laboratory, where a 2.14 m section of its length was documented. It seems that it extended more to the north-west, but no wood remains survived there. To the south-east an additional fragment of c.200 mm was photographed under water. This fragment was not taken to the laboratory, as it was detached at a point where the iron concretion of a bolt, originally connecting the CLT to the keel, had disintegrated.

Figure 12.

CLT a) under water (two photographs superimposed, above) (A. Yurman); b) drawing. (R. Gilboa)

The CLT was between 80 and 110 mm sided, 100 on average, and between 60 and 170 mm moulded, 150 mm on average. It was slightly curved, crudely worked, and bolted to the spine of the hull, through or between frames. Remains of a nail were found on the upper surface of the CLT section in the laboratory, 250 mm from its original south-eastern end, at about the location of the transverse bulkhead. Apparently, the bulkhead was connected to the CLT at this point. The CLT was recessed along its underside for fitting framing timbers. These recesses were cut 50–70 mm into the underside of the CLT: a single recess for floor timbers and a double recess for two overlapping half-frames.


Fifty-nine extant framing timbers and frame fragments were found in 28 frame stations (Fig. 13). Eight additional frame stations where no wood survived were identified, based on nail remains and staining patterns. Framing timbers were floor timbers, half-frames and futtocks. The majority were found in a poor state of preservation, eroded, broken, and some not in their original positions. At the north-western section of the hull several frames survived in their entirety, as did a few frames about the centre of the shipwreck, while in the south-eastern section of the hull only a few framing timbers survived. Near the north-western end, from the north-eastern side, many framing timbers were found closely spaced, touching each other, as a result of the convergence of the hull towards the end and the overlap of the floor timbers with the futtocks. Fewer framing timbers were found on the south-western side. The frame dimensions varied in width, thickness and space: they were between 50 and 130 mm sided, 100 mm on average; and between 20 and 140 mm moulded, 120 mm on average. Room and space dimensions were measured where accessible. They varied between 180 and 380 mm, with an average of 260 mm.

Figure 13.

Framing pattern and frame tree species. (E. Israeli, H. Itzkovitch)

About the centre of the shipwreck, a few frames were preserved in good condition (half-frames, FW26, FE27, Fig. 14), including the widest intact frame station containing wood, 2.2 m long. These half-frames were each connected to the keel by a single iron nail, but were not connected to each other. They apparently formed the midships frame. To the north-west (FW24, Fig. 13) and south-east (FW27) of the midships frame, floor timbers were found. From half-frames FW26 and FE27, towards the ends of the hull, the futtocks were located at the side of the frame facing the end of the hull. The joints between pairs of half-frames also changed direction from this point: towards both ends the north-eastern half-frames are on the side facing the centre and the south-western half-frames are on the side facing the ends.

Figure 14.

a) Half-frames FW26 and FE27, and floor timber FW27-FE28, an overview (A. Yurman); b) cross-section at FW26 and FE27 (below). (E. Israeli)

The framing pattern, with a few exceptions, was of alternating floor timbers and half-frames. Most of the frames were fixed to the keel by one iron nail with a square cross-section of 7 × 7 mm. Floor timbers were connected at their centre, and half-frames c.20 mm out from the centre line. In four frame stations no remains of frame-keel nails were identified (FE12, FE26, FE33, and FE37).

Limber holes were identified in the majority of the surviving frames. They were triangular, and roughly made. The limber holes were on average 20–30 mm wide and deep (vertically), and started 30 mm from the keel (Fig. 15).

Figure 15.

A limber hole in FW26. (I. Grinberg)

Frames had various cross-sections (Fig. 16). Some were roughly shaped (Fig. 16.1, FE23), and some were made of partially worked half-logs (Fig. 16.2, FW24). Several frames, mainly those at the centre of the shipwreck, were intensively worked, and had a rectangular cross-section with chamfered corners (Fig. 16.3, FE22, FE26, FW27, Fig. 17). A few frames had a trapezoidal cross-section, with the narrow side down (Fig. 16.4, FE24, FE25); and some had a rectangular cross-section, set horizontally with the wide side down (Fig. 16.5, FW27). This is not the optimal method of reinforcing the hull. The variety of frame shapes may indicate a shortage of timber, or a shipwright without many resources, who, although skilful, invested only a minimal effort in building the vessel.

Figure 16.

Frame cross-sections, a reconstruction. (E. Israeli)

Figure 17.

Frame FE22. (A. Yurman)

The frames were made of seven different tree species (Fig. 13). The majority were of Pinus brutia, and the others of Tamarix (X5), (X4), Quercus cerris, Quercus coccifera, Ulmus campestris and Fraxinus excelsior (Liphschitz, 2006; Liphschitz, 2007a; Liphschitz, 2008; Liphschitz, 2010). This variety in frame shape, cross-section and tree species may also indicate many years of service and many repairs.


Sections of 23 strakes made of Pinus brutia and Cupressus sempervirens survived, 13 on the south-western side of the keel, and 10 on the north-eastern side (Fig. 18). The longest plank was 3.1 m long; it was a section of the south-western garboard (PLW1/GBW), which survived to a total length of 6 m. The planks were a maximum of 200 mm wide, some as narrow as 40 mm, and were between 19 and 29 mm thick, 24 mm on average.

Figure 18.

The hull planking, repairs (yellow) and charring remains (shaded). (E. Israeli)

The majority of the narrow planks were repairs, all made of Cupressus sempervirens. The planks were connected to the frames by one to three iron nails, depending on the plank width. Nail remains show that the garboards were nailed into rabbets in the keel and the gripe. These iron nails had an average square cross-section of 7 × 7 mm. Planks were butt-jointed at frame stations (Fig. 19).

Figure 19.

Planking butt-joint at frame station (note nail remains). (A. Yurman)

No evidence of any kind of planking edge-joints was identified. The planks were irregularly shaped, exceptionally short, and some were very narrow, whether due to a shortage of suitable timber, or as a result of many years of service and repairs. There is no doubt that this planking shell was based on frames, and could have never been of any type of a shell-first hull. Since the surviving planks were in a very bad condition, and visually confusing, red string was used to mark the seams (Fig. 20). This technique was found to be very effective, combined with underwater photography and the creation of a photomosaic, for clarifying the planking pattern.

Figure 20.

Planking with coloured string in seams. (A. Yurman)

Examination of the documented planking pattern indicates many plank replacements or repairs. These small graving pieces (Steffy, 1994: 291) were butt-jointed between two, three or four frames (Fig. 18).

Caulking material was also used for repairs. The largest occurrence was 150 mm long and 20 mm wide. Remains of caulking material were found in many planking seams, in the keel rabbets, and in plank damage repairs. The caulking material contained natural fibre, which has so far not been identified.

Charring was found on many plank surfaces, sometimes a few millimetres thick. Fire could have served for bending the timbers, or against woodworm and pests. As it was generally found on plank ends, sometimes under frames, and on stringers, but not on higher timbers such as frames and ceiling planks, it may indicate the controlled use of fire. Documentation of the planks containing charring remains was complicated. It shows that the majority of the planks north-west of the transverse bulkhead were charred, while those on the south-east were generally not charred, perhaps as the result of wear and tear (Fig. 18). None of the small repair planks were charred, indicating that the repairs were made after the general application of controlled fire to the hull.

Remains of yellow-brownish pitch were found on the majority of the internal surfaces of planks. It was clearly discernible, mainly in frame-plank matching lines. Where frames were removed from their original location, no pitch remains were found. In places where pitch was applied on charred surfaces, it tended to peel. Incised lines were found on plank surfaces, indicating frame locations (Fig. 21). These lines were probably made during hull repairs, in order to record the position of the frame before removing it and replacing it precisely in its previous position. It seems that this hull underwent several major repairs. The positions of frames intended for replacement were incised in the planking, the frame removed, fire was used on infected planking surfaces, and frames were replaced in their original positions.

Figure 21.

Planks with incised lines and paint pitch remains over charring. (A. Yurman)


Sections of eight stringers were found (Fig. 22, Table 2): five on the north-eastern side of the keel (StE1, 2, 3, 4, and 6—StE5 did not survive), and three on the south-western side (StW1, 2 and 6—StW3, 4, and 5 did not survive). As they were located relatively high in the shipwreck, they were badly preserved, and were found broken into short sections, with decayed ends. The longest stringer timber, StE4, was 3.24 m long. Additional fragments belonging to this stringer made its total length 3.94 m. The lower stringers were better preserved. The stringers were made of Cupressus sempervirens, except for stringer StE1, which was made of Pinus brutia.

Figure 22.

a) Stringers on the north-eastern side (A. Yurman); b) and c) reconstructions of typical cross-sections (dimensions in mm). (E. Israeli)

Table 2. Stringer details
StringerSurviving length (mm)Maximum width (mm)Maximum thickness (mm)Tree speciesNail details
StE1257014652Pinus brutia4 nails, Ø 10 mm
StE230758040Cupressus sempervirens
StE35807020Cupressus sempervirens
StE4394015050Cupressus sempervirens2 nails, Ø 7–8 mm
StE6 (StE5 absent)7908530Cupressus sempervirens
StW1155014560Cupressus sempervirens2 nails, Ø 10 mm
StW2 (?)15207676Cupressus sempervirens
StW6 (StW3, 4, and 5 absent)3206534Cupressus sempervirens

The stringers were manufactured from half-logs, and were installed on the frames with their flat side down. Shallow recesses, made by adze, were found in the underside of the stringers, for fitting onto the frames. Some of the frames were worked on their upper surface for fitting the stringers. The stringers were worked flat on their upper surface. Based on their dimensions they were divided into two groups: the larger stringers, StE1, StW1, and StE4 were on average 147 mm wide and 54 mm thick (Fig. 22c); and the smaller stringers, which were on average 75 mm wide and 30 mm thick (StW2 not included) (Fig. 22b).

Signs of charring were evident on the upper surfaces of StE1, StW1 and StE4. Apparently these markings are the result of bending through a heating process, known as ‘char-bending’ (Greenhill, 1957: 115). The north-western end of StE1 was trimmed to fit the other converging stringers at this end where they touched the CLT. Timber StW2 had a round cross-section, with maximum diameter of 76 mm. No nail remains were evident in this timber. As it was found above the ceiling planks, it may not have been a stringer, but rather a piece of an oar or a spar.

The stringers were connected to the frames by iron nails 0.50–1.20 m apart (spaced between three and six frame stations). The nails left 10 mm-diameter holes in the stringers. Based on the north-eastern stringers, in addition to strengthening the hull, the large stringers served to support the ceiling planks (Fig. 23). The ceiling planks were installed between two large stringers, and were supported from below by the small stringers.

Figure 23.

Cross-section of the stringers and ceiling planks on the north-eastern side, a reconstruction. C2/10, C12 and C13 are adjacent to each other. (E. Israeli)

Bulkhead support and stanchion

The remains of a base for a transverse bulkhead made of Pinus brutia were found 2.75 m from the north-western end of the shipwreck (Fig. 24). It was supported on the stringers and the CLT, and fixed to them by one iron nail to each stringer and the CLT. The bulkhead base was made of two timbers that were found broken into several fragments. The total length of the archaeological remains was 2.10 m, and it crossed parallel to F24 and above it, between StE4 and the position of the missing StW4. It had a cross-section of 95 × 95 mm. A rectangular groove, 25 mm deep and 17–25 mm wide, was cut in its upper surface.

Figure 24.

The bulkhead base and the stanchion remains. (I. Grinberg)

About 260 mm from the inner ends of the two sections of the bulkhead base, slots 90 mm long and c.60 mm wide penetrated the entire thickness of the timbers. They were located above mortises in stringers StE1 and StW1, which were 90 × 50 mm and 70 × 35 mm respectively. These slots and mortises were bases for stanchions, a fragment of one of which was found in situ.

The underside of the eastern section of the bulkhead support had a circular recess, 200 mm long and maximum 60 mm deep, to fit over StE4. At its eastern end it was cut diagonally, 55° from the upper surface of the timber, corresponding to the angle of the turn of the bilge. The end of the south-western section did not survive.

Remains of a stanchion were found in the south-western section of the bulkhead base. It was 125 mm long (high), with a rectangular cross-section 72 × 68 mm. Its lower end was shaped as a tenon 50 mm long, with a rectangular cross-section of 30 × 25 mm, for fitting into the mortise in StW1. No similar stanchion survived in the recess in the north-eastern section of the bulkhead base or in the mortise in StE1. The existence of a bulkhead is a hypothetical reconstruction based on the archaeological finds of the base, the stanchion and other features mentioned above (Fig. 25). It is similar to the bulkheads surviving in the shipwrecks of Yenikapı 3 and Yenikapı 12 (Kocabaş 2008: 152–5).

Figure 25.

Reconstruction of the bulkhead, view from the NW. (E. Israeli)

Ceiling planks

Thirteen fragments of ceiling planks were found, comprising 11 planks (Fig. 26, Table 3). Several of the better-preserved ceiling planks were retrieved for study in the laboratory. Their maximum length was 912 mm, maximum width 272 mm (minimum is meaningless, as some were broken fragments), and they were between 23 and 33 mm thick. They were made of Cupressus sempervirens, except for C2, C10, and C12, which were of Pinus brutia.

Figure 26.

The ceiling planks in situ. (R. Gilboa and E. Israeli)

Table 3. Details of ceiling planking
Ceiling plankLength (mm)Width (mm)Thickness (mm)Nails (Number, diameter [Ø] in mm)InscriptionsTree speciesNotes
C176320424Cupressus sempervirensBroken lengthwise
C2315 (Broken)115 (Broken)29Δ ΔPinus brutiaFragment of C10
C387221929One nail, Ø 10.2ΔCupressus sempervirens
C490210727One nail, Ø 10.2Cupressus sempervirensBroken into two sections
C580721923Three linesCupressus sempervirensTrapezoidal recess
C669817729Five nails, Ø 10.0–10.4Cupressus sempervirensTrapezoidal recess, saw marks
C780720824One nail, Ø 7.0Two linesCupressus sempervirensTrapezoidal recess
C874320424Cupressus sempervirensAdze marks
C10522 (Broken)27233ΔPinus brutiaContinuation of C2, adze marks
C1175721225X A ΔCupressus sempervirensBroken lengthwise
C129129827Pinus brutiaTwo fragments
C1384323627Two nails (not measurable) image I Γ, ACupressus sempervirensBroken lengthwise

From the bulkhead base towards the north-west, the planks were installed longitudinally, and from the bulkhead towards the south-east they were installed transversely (except C4). Three ceiling planks had a trapezoidal notch in one of their edges. Saw and adze marks were found on some ceiling planks. They were shaped for fitting to the supporting stringers, both on their undersides and their edges.

Seven ceiling planks were rectangular (c.800 × 220 mm). Two (C7 and C6) were trapezoidal, apparently in order to correspond to the convergence of the hull. C12, which was located between C13 and the bulkhead base timber, was exceptionally long and narrow (912 × 98 × 27 mm), and was probably a filler. Similarly C4, between stringers StE4 and StE6, and assumed preliminarily to be stringer StE5, is probably also a filler (Figs 23, 26).

Nail remains were found at the ends of five ceiling planks: C3, C4, C6, C7 and C13. Three nail holes, 8–10 mm in diameter were found at the end of C6. This means that more than half of the ceiling planks were not connected to the stringers or the frames. They could be lifted by hand, in order to gain access to the bilge.

On C5 three, and on C7 two, straight transverse parallel lines were incised, c.130–160 mm apart. Greek letters were found incised in five of the ceiling planks: C2 – ‘Δ Δ’; C3 – ‘Δ’; C10 – ‘Δ’; C11 – ‘X A Δ’; and C13 – ‘image I Γ’ and a single ‘A’ (Fig. 27). All the letters were identified as Greek, typical of the period of the shipwreck (Kritzas pers. comm., 2008). In Tantura B similar inscriptions were found with straight lines and the letters ‘XH’ (Kahanov et al. 2004: 122). The meaning and purpose of these inscriptions is as yet unknown.

Figure 27.

Greek letters incised in the ceiling planks. (R. Gilboa)

Among and above the longitudinal ceiling planks, adjacent to the CLT from the north-east, was a long, loose, narrow, broken timber (778 × 88 × 29 mm). A single pegged mortise-and-tenon joint was identified at its centre (Fig. 26), the only example found in the hull remains. This timber was thinner and narrower than other ceiling planks.14C analysis dated it 200 years earlier than the shipwreck (see below). The plank, tenon and tapered peg were all made of Cupressus sempervirens. It was probably not part of this hull, and could have been deposited in the shipwreck by the waves and currents in the lagoon.

Tree species

Wood samples for species identification were taken from each hull component. The hull was made of seven different tree species: Cypress (Cupressus sempervirens), Calabrian pine (Pinus brutia), Tamarisk (Tamarix spp.), Turkey oak (Quercus cerris), Kermes oak (Quercus coccifera), Common elm (Ulmus campestris) and Common ash (Fraxinus excelsior). All tree species were identified by Prof. Nili Liphschitz of Tel Aviv University (Liphschitz, 2006; Liphschitz, 2007a; Liphschitz, 2008; Liphschitz, 2010) (Table 4). The possible geographical growing areas of the different tree species are shown in Table 5. The dominant tree species used in the hull were Cupressus sempervirens (42.9% of the samples), Pinus brutia (31.3% of the samples) and Tamarix spp. (20.5% of the samples). These three species grow together in Syria, Turkey, and Cyprus. Therefore, it is suggested that the trees from which the ship was constructed originated in Syria, Turkey, or Cyprus. However, they could have been imported to a shipyard in the Levant or Egypt.

Table 4. Tree species of hull components
 Longitudinal spine (keel etc.)FramesStringersHull planksCeiling planksPlanking repairs
Cupressus sempervirens+++++
Pinus brutia++++
Tamarix spp.++ 
Fraxinus excelsior+
Quercus cerris+
Quercus coccifera+
Ulmus campestris+
Table 5. Growing areas of the tree species used in the hull. Data based on Meusel, 1964; Coode and Cullen, 1965; Davis, 1965; Yaltirik, 1965; Renfrew, 1973; Zohary, 1973; Baum, 1979; Meiggs, 1982; Liphschitz, 2004: 157, 2007b
 SyriaTurkeyGreece (mainland)CyprusAegean islandsBalkansItaly
Cupressus sempervirens++++++
Pinus brutia+++++
Tamarix spp.+++
Fraxinus excelsior+++++
Quercus cerris+++++
Quercus coccifera++++
Ulmus campestris+++++

Carpenters’ toolmarks

Saw, adze and gouge marks were found on many hull components. Drilled holes were also found. Saw marks were found on the longitudinal spine components (the keel, the gripe and the connecting timber), the CLT, hull planks, ceiling planks, stringers, and on the sides of several well-preserved frames about the centre of the hull. The saw lines are at an angle to the width of the component, increasing as the width of the timber decreases: wide planks (150–250 mm) bore lines at an angle of 10°–12°; the keel (170 mm moulded) at an angle of 15°; and narrow (50 mm) repair planks, at an angle of 22°. On the keel, crossing saw lines were observed, which may be considered as evidence of the use of a two-man saw. The space between the saw lines varied between 2 and 8 mm.

An adze with a flat blade, 30 to 50 mm wide, was used on the lower surface of the stringers and ceiling planks for fine shaping (Fig. 28). These adze marks were spaced at c.5–10 mm. A gouge, 15 to 25 mm wide was used to create the longitudinal grooves in the bulkhead base. These marks were spaced at c.5–20 mm (Fig. 29).

Figure 28.

Adze marks on the lower surface of ceiling plank C13. (J. J. Gottlieb)

Figure 29.

Gouge marks in the groove of the bulkhead base. (R. Gilboa)

Drilled holes, 14–16 mm in diameter, were evident in the timbers of the longitudinal spine and the CLT. Bolts for securing these timbers were driven through these holes. The longest drilling found was 170 mm. A drilled hole, 15 mm in diameter, was found in ceiling plank C6. It was well preserved, and the bit cuttings are still visible.


Iron fasteners of various dimensions were used to connect hull components. Although the metal hardly survived, their dimensions can be estimated by a few scarce metal remains, nail holes, concretions and several nail-head indents seen in the timbers. Iron concretion was found in three scarfs of the longitudinal spine (S2, S3, and S4). Drilled holes of similar dimensions were evident in the endpost fragment, the gripe and the CLT. As the diameter of the holes was c.16 mm, without nail scars inside the hole, and there is evidence of a square nail (Fig. 30), it seems that iron bolts with a square cross-section 11.5 mm (calculated), and a round head 35–50 mm in diameter were used. According to the timber thicknesses their length was c.200 mm long (average moulded 150 mm, and securing the end 50 mm), and the CLT–keel bolt was c.450 mm long (combined average moulded of the CLT [150 mm minus 20 mm of underside recess], frames [120 mm], the keel [150 mm], and securing the end [50 mm]).

Figure 30.

Square keel bolt remains. (A. Yurman)

Frames were connected to the longitudinal spine of the hull by square iron nails, one nail per framing timber. These nails tapered from near their heads, where they were on average 9.5 × 9.5 mm in cross-section, to a sharp point. Their heads were round, 20–25 mm in diameter.

Traces of five nails that connected the wide stringers (StW1, StE1, and StE4) to the frames were found. They tapered from a square cross-section, c.9.5 × 9.5 mm near the head, to a sharp point, and had round heads 20–25 mm in diameter. Their length can be estimated at c.160 mm (54 mm average stringer thickness plus double this measurement inside the frame).

The majority of the recorded nails connected hull planks to frames—one to three nails per plank per frame. They were iron nails, with a tapered cross-section c.7 × 7 mm near the head, and a round flat head 15–20 mm in diameter. Their length can be estimated at c.70 mm (24 mm average plank thickness plus double this measurement inside the frame). Similar nails, perhaps slightly longer (c.90 mm), connected the narrow stringers to the frames from above.


The in situ finds included stones, broken bricks, a ceramic disc and amphora sherds.

Nine stones were found above the ceiling planks, designated St1–St9. Two ashlars (St1 and St2) of coarse calcareous sandstone (locally known as kurkar), were found side-by-side, above the ceiling planks, and closely fitting along the transverse bulkhead base. The weight of St1 was c.28 kg, and of St2 c.42 kg. The locations of these stones and the groove in the underside of St1 that fitted stringer StE4 probably indicate that the two stones were part of the ship's ballast, most of which was not found (Fig. 31).

Figure 31.

Stones found: a) St1 and St2 in situ (I. Grinberg); b) St1 (R. Gilboa)

In the south-eastern part of the wreck, close to the south-western side of the keel and above the hull planks (where the frames and ceiling planks did not survive), a cluster of brick fragments and a ceramic disc were found. The bricks were c.157 mm long (only one complete brick survived) 74–89 mm wide (six bricks) and 43–57 mm thick (eight bricks). They were of a light red colour—2.5YR5/2 (Munsell, 1994) with white inclusions. NAA analyses indicate that the origin of the raw material was Egypt (Mommsen, 2012) (Fig. 32).

Figure 32.

a) Brick remains recovered (E. Israeli); b) bricks 1 and 7 (right). (R. Gilboa)

The ceramic disc was 212 mm in diameter and 42 mm thick, and was a light reddish-grey colour (5YR6/4) with brown inclusions. NAA of the disc attributes the origin of the raw material to Cyprus (Mommsen, 2012) (Fig. 33). Similar discs were used to support hypocaust floors during the Roman and Byzantine eras, such as in the well-preserved bathhouse in Beth-She'an (Atrash, 2006: 37, fig. 88). White remains of bonding material were found on one face of the disc, and the other face was charred. Combined with the small quantity of bricks, it might be suggested that these are the remains of a simple onboard cooking stove.

Figure 33.

Ceramic disc: a) in situ (A. Yurman); b) photograph (J. J. Gottlieb); c) drawing. (R. Gilboa)

Fragments of a single whole amphora, B0013 (Fig. 34), were found in the north-western part of the wreck, above the ceiling planks, close to the north-western side of the bulkhead base. The outsides of the ceramic fragments were yellowish-red (10YR6/1) or grey (5YR7/6), with a white-painted spiral and parallel vertical lines; and grey (10YR5/1) with remains of pitch on the inside. They were well fired. This vessel, 492 mm high and 367 mm in diameter, with a volume of c.35 litres, is a typical local ‘bag-shaped’ amphora of the local Umayyad period (641–750 CE). Petrographic analysis indicates that the origin of the raw material was the eastern Mediterranean, perhaps the Aegean islands or Cyprus. This amphora has several parallels from Pella (Smith, 1973: 234), Jerash (Clark, 1986: 335), and Hippos (Sussita) (Segal et al., 2003: 71). In these sites, the parallels were found under the destruction rubble caused by documented major earthquake events in the Jordan Valley and trans-Jordan of 748 or 749 CE.

Figure 34.

Bag-shaped amphora B0013 after restoration, a) photograph (J. J. Gottlieb); b) drawing. (R. Pollak)

A cluster of 19 lead fishing-net weights (W1–W19) was found at the centre of the site, above the ceiling planks. All the weights but one were of the double-folded rectangular sheet type (Galili et al., 2002; Galili et al., 2010). A herringbone-motif decoration can be identified on three of the weights. The weights were found together, and probably belonged to one net. However, is not certain that they were in situ, and part of the ship's equipment.

Dating of the shipwreck

Nine organic items, some of short-living materials, were 14C-tested. Their calibrated dates were mid 5th–early 9th centuries CE (1SD)/mid 5th–mid 9th centuries (2SD) (Fig. 35). The calibrated ranges were obtained using OxCal 3.10 (Bronk Ramsey, 2005), based on the calibration curve of Reimer et al. (2004). A short-living sample (ETH-32880) taken from the remains of a rope, was dated to 670–780 CE (1SD)/650–870 CE (2SD). Based on the 14C results and the typology of the amphora, the date of the shipwreck is suggested to have been between the 7th and 9th centuries CE—spanning the end of the Byzantine to the early Islamic periods in the region.

Figure 35.

14C results


The recording methods of Tantura E included direct measurements, photography, photomosaic, hand-drawing and a direct survey method (DSM or WEB [Rule, 1989]).

Several cross-sections and a longitudinal section of the keel were generated. In addition, significant components, among them several frames, the endpost fragment (K1), the gripe (K2), the connecting timber (K3), and the CLT were retrieved from the sea-bed, and were scanned by high-resolution laser technology, allowing the use of CAD programs which facilitated extracting dimensions and cross-sections.

The cross- and longitudinal-sections made on site allowed the partial reconstruction of the shape of the archaeological find. Integration of these with laser scans, WEB measurements and other records, allowed locating and connecting virtually, step-by-step, all surviving keel members and some of the frames, and can be shown in three-dimensional views (Fig. 36). This integrated data led to the preliminary understanding of the original dimensions at the very lowest part of the hull, and was the base of the suggested hull lines plan (Fig. 37).

Figure 36.

3D reconstruction of the longitudinal spine and cross-sections, view from the north (the numbers 165–338 are distances from the north-western end of the shipwreck in cm). (E. Israeli)

Figure 37.

Suggested hull lines for Tantura E. Bow to right. (A. Ben Zeev, E. Israeli)

The archaeological information gave a reasonable indication as to the bottom of the hull: transversally up to the turn of the bilge and longitudinally from one endpost at the north-west 7.6 m towards the south-east, to the end of the hull remains. Where the archaeological evidence ended, the reconstruction is based on the following information: contemporary shipwrecks, such as Tantura F for the longitudinal symmetry (Barkai and Kahanov, 2007), Yenikapı 17 for the two wales (Kocabaş, 2008: 170), Yenikapı 3 and Yenikapı 12 for the bulkhead location, which indicates that the north-western end of Tantura E was the stern (Kocabaş 2008: 112–124, 152–163); contemporary visual evidence, such as Skylitzes’ ‘Sinopsis Historian’ illustrations (Codex Skylitzes Matritensis, Bibliteca Nacional de Madrid, Vitr. 26–2, Bild-Nr. 77, f 34 v. b.); the ships from the Damascus mosque (Nicolle, 1989: 170, fig. 4); the ship illustration from the St Gregory of Nazianzus manuscript for the reconstruction of the upper section of the hull and endposts (Landström, 1961: 81, figs 209, 210; Pryor, 1988: 26, fig. 4; Pryor and Jeffreys, 2006: fig. 15); and traditionally built vessels from the eastern Mediterranean for hydrostatic characteristics and waterline estimation (Traung, 1960; Damianidis, 1999: 225, 226 fig. 1, 279 fig. 20).

Based on the above information the preliminary suggested dimensions of Tantura E are: length overall (LOA) 12.5 m, beam (Bm) 4 m and depth (D) 2.3 m (Fig. 37). The suggested draught/depth ratio (T/D) is 3:5 (Stone, 1946; Traung, 1960; McGrail, 1987: 13; Tupper, 1996; Damianidis, 1999: 225, 226 fig. 1, 279 fig. 20). Assuming a similar ratio in Tantura E, the suggested draught (T) was 1.4 m. Using Nautilus software, the hydrostatic properties were calculated, giving a displacement corresponding to 1.4 m draught of 25 tonnes.

According to the above reconstruction the unladen weight of Tantura E with four crew members is estimated at 7.5 tonnes, with vertical centre of gravity (VCG) c.0.8 m above the keel. In this case the ship could have loaded 17.5 tonnes of cargo and ballast.

Stability was cross-checked with two modern regulations: UK Marine and Coastguard Agency MGN 280 (M) for small vessels in commercial use, and Register Holland: Rules for seagoing vessels with length between 12–40 m Lpp up to 36 passengers, both in common use for small modern vessels. According to Register Holland, basic stability for Tantura E could not be achieved only with cargo, and significant use of ballast was required, such as a single layer of stones above the ceiling planks. Three tonnes of ballast and 14.5 tonnes of cargo provide a centre of gravity at 1.1 m above the keel.

A sail was also tested. Two basic assumptions were made: 1) The general area of the sail was about the same as the wetted area (Matthews, 2004: 179–185; Ben Zeev et al., 2009: 68); and 2) A lateen sail was assumed (Pryor, 1988: 27; Polzer, 2008: 246). The wetted area of the proposed reconstructed hull, at a draught of 1.4 m is 41.5 m2; thus one lateen sail, of area c.42 m2 is suggested. Tests based on Register Holland showed that Tantura E could sail safely up to Beaufort 5 wind conditions (and Beaufort 6 gusts), and survive with the sail furled in any wind condition recorded in the eastern Mediterranean (Mediterranean Pilot V: 29 diagram 1.151.1–32 diagram 1.151.4).

The complete midship cross-section was reconstructed containing two half-log wales (Fig. 38, nos 10, 11), separated by a 200 mm hull plank, two additional stringers (Fig. 38, nos 8, 9) of the same dimensions and spacing as found in the surviving lower areas (Fig. 38, nos 3–6), and homogeneous hull planking with thickness of 25 mm (Fig. 38, no. 2).

Figure 38.

Schematic reconstruction of the hull cross-section. (E. Israeli)

Transverse components, such as frames and ceiling planks that have no longitudinal strength, were ignored. Longitudinal hull strength was calculated, assuming the hull to be a loaded beam (White, 1979). Under the strict assumption of loading case, assuming that the hull is subjected to heavy loads at the fore and aft ends, shear forces and bending moments were calculated—maximum shear force 3 tonnes, 2 m from both ends, bending moment 9.5 tonne-m at midships. The results show that the keel and bottom planking, the most highly tensioned components of the hull, are less stressed than the maximum tension allowed for pine and cypress timbers. (Maximum tested tension in Anatolian pine along the grain: σ = 270 kg/cm2 [Aydin et al. 2007: 23]. Maximum tension for Anatolian cypress along the grain: σ = 418 kg/cm2 [Bektaş and Kurt, 2010: 359].) This may be the builder's allowance for long years of wear, teredo damage and marine environmental effects.


The most obvious characteristic of the hull of Tantura E is its construction from many short planks, combined with many small timbers, apparently repairs. These planks were dependent on being connected to pre-existing frames by nails. The fact that no planking edge-fasteners were used, planking seams were caulked, planks were joined to form strakes by butt-joints at frame stations, and other criteria, are clear proof that this hull was built based on frames. There are a few plank diagonal joints, which scholars may suggest are indication of shell-first construction. These joints are without edge-fasteners, were found without charring remains, unlike the adjacent planks, and are of a short length. Thus, these small planks could have been repairs. Therefore, considering all other construction evidence as described above, we feel that these diagonal planks are not evidence of shell-first construction.

The connection of the garboards to the keel may sometimes help in distinguishing between shell-first (garboard connected to the keel) and frame-based (garboard not connected to the keel) (such as Dor 2001/1, Mor and Kahanov, 2006: 275, 288), but it is not a strict distinction, and there are hulls where the garboards are fastened to the keel rabbet only with nails and no edge-fasteners. Tantura E was based on frames; however, the garboards were nailed to the keel in a keel rabbet.

Slightly north-west of the centre of the shipwreck, the measured angle of deadrise was 6°. About the centre of the shipwreck the frames were almost flat, with a hard chine (Figs 37, 38). The construction tradition of Tantura E, based on the cross-section has been discussed by Pomey et al. (2012: 304). Combined with the Egyptian origin of the clay bricks, a Nilotic origin of the construction tradition cannot be ruled out. The majority of tree species used in the hull were from western Turkey, Syria or Cyprus. The timber could have been transported to the shipyard, or the shipyard could have been located anywhere along the Levant coast where the Nilotic tradition had spread and survived, subject to local variations, although the origin of the raw material of the bricks points towards Egypt.

Dor/Tantura lagoon is a dangerous, shallow, narrow and tricky anchorage. It is only partially protected by four small islands (La Jonquière, 1900: IV, 379; Kahanov, 2011: 169, 178). There is a narrow navigational channel inside the lagoon, where a current that can reach 2–3 knots sets from north to south. The prevailing winds are from the western quarter during the day and a light easterly breeze during the night. Thus, a ship entering the lagoon, must turn northward, may lose the wind, and without steerage way, face a counter current, which would drive it ashore. Thus local knowledge is required in order to enter and anchor safely. Safe anchorage in the lagoon is limited to wind strength of Beaufort 4–5, and wave height of 1.5 m (Kemp, 1976: 71–2), above which, staying inside the lagoon may be fatal. It would be used only in good weather and for specific purposes, such as rest for fishermen or ships’ crew, loading supplies, or commerce with known local people.


Tantura E was a small coaster or a fishing boat, dated to between the 7th and 9th centuries CE. The hull was built of eastern Mediterranean tree species. Its estimated dimensions were: length overall 12.5 m, beam 4 m, and depth 2.3 m. At a draught of 1.4 m the ship had a displacement of 25 tonnes. The carpentry was strong and solid, with considerable allowance for wear and tear for many years of service. It was constructed with a rigid keel and stringers and a relatively dense framing pattern, but the planks were exceptionally short, with many repairs, whether due to a shortage of suitable timber, or as the result of many years of service and repairs. There were no planking edge-joints, and the seams were caulked, indicating that the hull was based on frames. Thus Tantura E is evidence of the completion of the transition in ship construction from shell-first to frame-based in this period. Its construction tradition might have been Nilotic. Fully loaded with up to 17.5 tonnes of cargo (and/or ballast) in proper and common practice (VCG 1.1 m) the ship had acceptable stability and sailing characteristics according to modern regulations, providing safe voyages in the Mediterranean under sail for almost the entire year, and could have survived under reefed or furled sail without any constraints in any local weather conditions.

Tantura E is also evidence of a sailing route along the Levant coast, and for the existence of a settlement in or around Dor in the local late Byzantine–early Islamic periods.


This research was supported by the Hecht Trust, a Sir Maurice Hatter Fellowship for Maritime Studies, and anonymous donors, to whom the authors are grateful. Thanks are due to the anonymous reviewers for their constructive comments, and to John Tresman for the English editing.