The Dor 2001/1 Shipwreck, Israel—a Summary of the Excavation

Authors


Abstract

Dor 2001/1 was probably a Byzantine coaster carrying building stones, dated to the end of the 5th or beginning of the 6th century AD. No shell-first construction features were identified, specifically no kind of planking edge-joints; but instead all the essential components showed that it was a frame-based hull. Thus it is apparently among the earliest frame-based shipwrecks found so far in the Mediterranean.

© 2006 The Authors

Dor (Tantura) lagoon is located on the Mediterranean coast of Israel, about 30 km south of Haifa. During a survey carried out by students of the Department of Maritime Civilizations of the University of Haifa in November 2001, fragments of wood of at least two different species surfaced, and were dated by 14C to the end of the 5th to beginning of the 6th century AD, which is the Byzantine period in this region. The shipwreck was named Dor 2001/1. The last of four excavation seasons, from 2002 until 2005, has now been completed. During the first season a general, though partial, idea of the wreck was achieved. In the second season the upper level of the cargo of stones was documented and removed (Fig. 1). During the third season the removal of the rest of the cargo was completed, and the wood remains were studied. In the fourth season the documentation and removal of the ceiling planks was completed, the bottom of the ship, mainly the frames, was studied, and a 2 m section of one side of the hull was transferred to the laboratory for further research and conservation.

Figure 1.

Top view of the hull with the layers of the cargo of stones. (C. Brandon, adapted by D. Gev and S. Haad)

The site and the hull remains

The shipwreck was oriented roughly north-west/south-east, about 70 m offshore, adjacent to the lagoon's navigation channel, at a depth of 1 m, and buried under 1.8 m of sand. The total length of the find was 11.5 m, and its maximum width was 4.5 m (Fig. 2). Several clues allowed extrapolation to the end of the ship, probably the stern, which did not survive. The other end, probably the bow, was found almost complete. Thus it was possible to estimate the length of the original ship at c.16 m, with a beam of c.5 m. The hull, at least amidships, was almost flat-bottomed, with the angle of deadrise increasing towards the bow and stern (Fig. 3). Flat-bottomed midships sections are not exceptional at this period (Steffy, 1994: 10–14).

Figure 2.

Top view of the hull, after removing the cargo. (S. Breitstein)

Figure 3.

Planking cross-section amidships (the remains of SE19 and wale 2 were retrieved from the sea before this was drawn). (C. Brandon)

The hull remains form about one-third of the original structure, and include sections of the keel, false keel, 40 frames (floor timbers, half-frames, and futtocks), 29 strakes, including a chine strake and two wales, 21 ceiling strakes, among them two foot-wales and a clamp, a central longitudinal timber, two central stringers, and a mast-step sister. Since the ship was not on an even keel, the north-eastern side was in a better state of preservation than the south-western side, and provided more information about the hull's cross-section (Mor, 2002–03; Mor, 2004; Mor, 2005; Kahanov and Mor, in press).

Wood analysis

Wood samples were taken from each hull component for species identification (Table 1). The hull was made of eight different species: Pinus brutia, Cupressus sempervirens, Quercus coccifera, Ulmus campestris, Quercus cerris, Fagus orientalis, Ziziphus spina christi and Tamarix X5. The origin of the first six was western Turkey, where they can all be found in the same area. The last two do not grow in western Turkey, but are local to the wreck-site (Liphschitz, 2004).

Table 1. Tree species and hull components (analysis by Prof. Nili Liphschitz, of the Institute of Archaeology, Tel Aviv University, and Mr Avner Hillman, of the University of Haifa)
Tree speciesHull components
Cupressus sempervirensKeel, planking (including the chine strake and 2 wales), central stringers, mast step sister
Fagus orientalisFrames
Pinus brutiaCeiling (including a clamp and foot wales), frames
Quercus cerrisFrames
Quercus cocciferaFalse keel, frames, central longitudinal timber
Tamarix X5Frames
Ulmus campestrisFrames, cleat in keel, endpost
Ziziphus spina christiFrames

Keel and false keel

The keel survived for almost 9 m, with at least one scarf about midships, where the section was made. It was made of softwood, Cupressus sempervirens, and was exposed for a length of 2.04 m towards its south-eastern end, while its upper surface was visible between the frames. It was 11 cm sided and 16 cm moulded (Fig. 4). There was no rabbet or chamfer for the garboards, which butted the keel from the sides, but were not connected to it. The seam between the garboard and the keel (as all other seams) was caulked. A rectangular notch was found 1.1 m from its south-eastern end (see Fig. 4). It was 7 × 8 cm, and penetrated the entire width of the keel. The notch was filled by a rectangular timber made of Ulmus campestris. No explanation of its purpose was found, though perhaps it could have served as a cleat when beaching or hauling the ship up a slipway.

Figure 4.

Keel and false keel. (S. Breitstein) Cross-shaped carpenter's marks and a cleat. (N. Sheizaf, adapted by S. Haad)

Remains of frame nails were observed on the upper surface of the keel at almost all frame stations. They were made of iron, with a rectangular section averaging 6 × 6 mm. The false keel was of hardwood, Quercus coccifera, and was 11–13 cm wide and 5–9 cm thick. Visible only along the exposed part of the keel, it was fixed to the keel by iron nails through pre-drilled holes at intervals of 20–30 cm.

Endpost

A poorly preserved endpost of hardwood, Ulmus campestris, was found lying at the north-western end of the hull. Its scarf with the keel was not observed. It was maximum 10 cm sided and 20 cm moulded (Fig. 5), and had a rabbet 9 cm beneath its upper surface, into which the planks on both sides were fitted, and fixed with iron nails to the endpost. The gaps were filled with caulking material.

Figure 5.

The northern endpost. (S. Breitstein)

Frames

Altogether, 42 frame stations were identified in the wreck, 40 of which were extant wood remains, and two were identified by nail remains on the upper surface of the keel. Most of the timbers survived on the north-eastern side of the wreck (Fig. 1). Floor timbers and half-frames were the pattern north of the mast-step sister (Fig. 6), between frame-stations E19a and E24a. In the mast-step area between frame stations E16 and E19 there were only floor-timbers. Floor-timbers spanned the bottom to the turn of the bilge, and were fastened to the keel with iron nails. Futtocks were diagonal-scarfed in the vertical plane to the floor-timbers with iron nails, and continued upward along the hull's side. In some places, a second futtock was found. Pairs of half-frames were scarfed above the keel and nailed to it. These half-frames were made of a single timber which formed the turn of the bilge (Fig. 7). At the end of the hull pairs of half-frames only were evident.

Figure 6.

Floor timbers and half-frames. (S. Breitstein)

Figure 7.

The turn of the bilge—half-frame and the chine strake. (S. Breitstein)

The floor timbers had average dimensions of 80 mm sided and 120 mm moulded; their futtocks had average dimensions of 85 mm sided and 90 mm moulded; the half-frames had average dimensions of 80 mm sided and 100 mm moulded. The average room and space was 24 cm. The frames generally had two limber-holes (Fig. 8), one on either side of the keel, above the garboards. They had average dimensions of 4 × 3 cm or 3 × 2 cm. Frames W12, W18 and W20 had three limber-holes, all about 4 × 3 cm. Frame W12 had a third limber hole above strake 3 on the north-eastern side; frame W18 had one above strake 2 on the south-western side; and frame W20 had one above strake 2 on the north-eastern side. Evidently, there was no alignment between the extra limber-holes. A possible explanation of their existence might be the use of prefabricated components, or a secondary use. The floor-timbers and half-frames were not notched for fitting onto the keel, but were fastened to the keel by iron nails, driven through pre-drilled holes. The nails were tapered, with a rectangular cross section. Their heads were generally not available for measuring as they were covered by concretion; but where accessible (on the keel surface), they averaged 6 × 6 mm.

Figure 8.

Limber-holes. (S. Breitstein)

The frames amidships had almost flat undersides at obtuse angles, and sharp turns at the bilge, but towards the bow and the stern they gradually converged with an increasing angle of deadrise. The hull was thus almost box-shaped amidships, where most of the cargo was concentrated, but the hull as a whole did not look like a barge. As seen in Table 1, the frame assemblage was made of seven different species of wood, most originating in western Turkey. However, the frames northward from E25 were mostly made of Ziziphus spina christi and Tamarix X5. It seems that they were installed locally, maybe as a repair after an accident, or as maintenance.

Planking

Sections of 29 strakes survived, all of Cupressus sempervirens: nine on the south-western side of the hull (SW1–SW9, SW1 being the garboard), and 20 on the north-eastern side (SE1–SE20, SE1 being the garboard). They were all attached to the frames by rectangular iron nails, with a square cross-section of 6–8 mm, which narrowed down to a 4 mm square section when they entered the frames. Where accessible, each plank-frame connection was found to be by two nails, about 3 cm in from the plank edge (Fig. 9). Most of the planks on the south-western side were in poor condition. Their widths varied from 100 to 170 mm, and their average thickness was 25–30 mm. In the centre of the hull the bottom on the north-eastern side was composed of 12 strakes, SE1–SE12. Thus the chine strake, which creates the turn of the bilge, was the 13th strake (SE13). It was made of a half-log, maximum 280 mm wide and 95 mm thick (Fig. 7). Above it, the strakes changed attitude from horizontal over the bottom to vertical, and were labelled SE14–SE20. SE18 was wale 1, made of a half-log, 210 mm wide and 105 mm thick (Fig. 10). Above it was an additional strake in a very poor state of preservation, and above that was wale 2 (SE20), also made of a half-log, 180 mm wide and 70 mm thick, which was almost detached and heavily damaged by teredo worms. The widths of the planks on the north-eastern side of the hull varied from 100 mm to 160 mm, and their average thickness was 25–30 mm. Two plank scarfs were identified in the form of a butt joint: one on the seventh strake (SW7) on the south-western side, where the two plank ends were nailed to frame W22 (Fig. 9); the other on the north-eastern side, two planks above the chine strake, under frame E26a.

Figure 9.

Butt-joint strake 7, south-western side. (S. Breitstein)

Figure 10.

Wale 1. (S. Breitstein)

Central longitudinal timber

The entire length of a central longitudinal timber of Quercus coccifera, 2.5 m long, 15 cm sided by 18 cm moulded, was found intact at the north-western end of the hull. It was initially identified as a keelson, but after the cargo was removed, the details became clear (Fig. 11). It was laid above the frames, with notches in its lower surface to fit them, and just touched the endpost itself (Fig. 5). It was attached to the keel and frames by five long nails, at least one of which spanned the gap between the lower surface of the central longitudinal timber and the keel. On its upper surface it had two large-diameter holes (see below).

Figure 11.

Central longitudinal timber and central stringers. (S. Breitstein)

Central stringers

There were two central stringers made of half-logs, of Cupressus sempervirens, one on either side of the central longitudinal timber (Fig. 11). The south-western central stringer survived for a length of 2.65 m, averaging 125 mm wide by 85 mm thick. The north-eastern central stringer survived for a length of 1.37 m, averaging 125 mm wide by 80 mm thick. They stretched towards midships, where they were broken. The gap between the ceiling strakes and the nail remains on the frames indicate that these two stringers probably continued the whole length of the hull. The space between the two central stringers amidships was probably partly occupied by the missing mast-step.

Mast-step assemblage

A timber 122 cm long, 35 cm wide and 7 cm thick, with a rectangular groove measuring 8 by 16 cm, lying at 90° to the hull's longitudinal axis, was probably one of the mast-step sisters (Fig. 2). Based on the evidence from Tantura F, the mast-step was attached to this component towards the bow (Barkai, 2005: 17–18). The mast-step was apparently located in the narrow longitudinal gap between the ceiling strakes and the two central stringers.

Ceiling

Twenty-one ceiling strakes survived on both sides of the keel, all of Pinus brutia. The 12 on the north-eastern side of the keel were labelled C1–C11, C1 being next to the north-eastern central stringer; and the nine on the south-western side were labelled CA–CI, CA being next to the south-western central stringer (Fig. 12). The ceiling planks were laid above the floor-timbers and half-frames, but not all were attached to them. Generally they alternated: a long plank attached to the frame beneath (C3, C5, C7 and C9 on the north-eastern side and CC, CE, CG and CI on the south-western side), and several short planks that were not attached to frames, rather simply laid upon them (C1+C2, C4, C6 and C8 on the north-eastern side and CA+CB, CD, CF and CH on the south-western side). The iron nails were rectangular, with an average cross-section of 5–7 mm, and were driven in on either side of the planks, above the frames, in no apparent pattern (Fig. 12). The remains of the nine ceiling strakes on the south-western side of the keel, CA–CI, lay horizontally above the frames, and survived in varying lengths, from 2.4 m (CH) to 4.4 m (CA). Their thickness was 20–25 mm, and their width varied from 135 to 200 mm, except for CA, which was 260 mm wide. Ceiling strake CC had a diagonal scarf with iron nails in both sides. CI was a 60-mm-thick foot-wale, and probably formed the turn of the bilge on the south-western side of the hull. Unfortunately, this is exactly where this side was broken, so the rest of its structure can only be assumed based on the construction of the hull's other side.

Figure 12.

Top view of the ceiling. (N. Yoselevich)

C1–C9 also lay horizontally above the frames. Just north of the mast-step sister their widths varied from 140 to 250 mm, and their thickness was 20–25 mm. South of the mast-step sister they had the same thickness, but they narrowed towards the southern end, using drop strakes, and their width varied from 60 to 90 mm. Some survived to a length of 3 m (C1) and some to 8.5 m (C5). C9 was also a 60-mm-thick foot-wale, like CI, installed at the turn of the bilge against the chine strake (SE13) (Fig. 7). Ceiling strake C10 and C11 changed attitude and were vertical, parallel to the ship's side. C10 averaged 170 mm wide, while C11, the widest ceiling plank that survived, averaged 270 mm, and 25–30 mm thick. It also had a long diagonal scarf attached by iron nails.

Clamp

A clamp, of Pinus brutia, was placed against wale 1, above ceiling strake C11, and was the 12th ceiling strake on the north-eastern side. Eleven cm wide and 6 cm thick, it survived to a length of 3.90 m. Unfortunately, because of its size, and after so many years under water, the removal of the cargo stones, which evidently had held it in place, caused it to disconnect from the nails that attached it to the frames (Fig. 13).

Figure 13.

The clamp. (S. Breitstein)

Tools and carpenter's marks

Saw-marks were identified on many frames, planks, and ceiling strakes. It is impossible to determine which type of saw was used, but there are differences between the marks themselves. In some places the saw-marks were at a diagonal angle to the plank itself, and the space between the tooth-marks was 4 mm. In other places, mostly on the frames, the saw-marks were perpendicular, and the tooth-marks were closer, spaced 2 mm apart. The limber-holes seem to have been made by sawing at their ends, the fill being broken out by angling the saw.

Two 2-cm-diameter holes, which seem to have been made by a drill or auger, were identified in the upper surface of the central longitudinal timber. Three cross-shaped (‘+’) signs were scribed in the western side of the keel, between 1.65 and 1.75 m from the south-eastern end of the keel close to the notch. Their average dimensions were 15 × 20 mm (Fig. 4).

The finds

About 80 stones were found in situ, apparently the main cargo of the ship. They were stacked in two layers of three to seven adjacent rows (Fig. 14). All the stones were slightly trapezoidal voussoirs of an arch, and were about the same size, with average dimensions of 57 × 28 × 18 cm, weighing about 45 kg each. They were laid on the ceiling planking, which was protected by a woven mat. The stones were of kurkar (coarse calcareous sandstone). They contain much thin-grained quartz, foraminifera and seaweed remains with carbonate fragments, which indicate that they originated from an inter-tidal zone. Petrographic analysis has ruled out their source as being Turkey, Cyprus and Syria; and they are probably local, but it is geologically impossible to determine a definite source, since for example, the local and Spanish material look very much alike (pers. comm. Mart, 2005). The stones were all measured and documented before they were removed layer by layer (Fig. 1), and placed in a specially-dug trench, 3 m west of the hull remains, about 2.80 m below sea level, and covered with sand.

Figure 14.

The cargo of building stones. (S. Breitstein)

The organic material included the mat on the ceiling planking (Fig. 15). Other than determining that it was made of broad leaves and gramineae, it was impossible to identify the species and origin (Liphschitz, 2002). A few pieces of rope were also exposed. They were dated by means of 14C (Table 2), but due to their poor state of preservation the material from which they were made has not yet been identified (Fig. 16).

Figure 15.

The mat. (S. Breitstein)

Table 2. 14C tests (organic materials were sent for 14C dating at the laboratories of the Weizmann Institute, Israel, and the accelerator at the Institute of Particle Physics, of the Swiss Federal Institute of Technology, Zürich)
Lab. no.Sample no.MaterialAge (y BP)δ13 [0/00]Calib. age (AD) confidence level
RT-4254Initial surveyWood1680 ± 60−27.8250–430
RT-4255Initial surveyWood1620 ± 50−23.7390–540
RT-4256Initial surveyWood1650 ± 60−26.5260–280 (5.6%)
320–530 (62.7%)
RT-4610G74-2002/1000Wood1520 ± 35−24.8430–490 (11.3%)
530–610 (56.9%)
ETH-25381Initial surveyWood1680 ± 50−25.8 ± 1.2249–459
ETH-26367G74-2002/1006Matting1625 ± 45−17.7 ± 1.1337–549 (99.8%)
ETH-28109G48-2003/2011Tree bark1590 ± 45−28.3 ± 1.1400–570 (96.2%)
571–594 (3.8%)
ETH-28110G48-2003/2010aRope1665 ± 40−14.6 ± 1.1260–290 (7.5%)
322–466 (83.2%)
473–532 (9.3%)
ETH-29913G64-2004/3015Rope1450 ± 40−26.7 ± 1.1541–659 (100.0%)
ETH-31268G55-2005/4006Tree bark1590 ± 45−23.8 ± 1.2385–583 (98.5%)
Figure 16.

A rope. (N. Sheizaf)

The site was rich in Byzantine sherds, which were found mostly above the wreck, some above and between the stones, and very few within the hull. Most of these were severely damaged, although some may be partially restored. They were mostly the remains of ‘Yassi Ada amphoras’, ‘Gaza Ware’ and Byzantine cooking pots (Figs 17 and 18). However, due to the nature of the lagoon, where the heavy winter storms move pottery and stones very easily, it is not certain that the ceramic remains belong to this ship (see Royal and Kahanov, 2005). Finds from the inside of the hull are of the same pottery types, but they are very few and mixed with Late Roman, Hellenistic and Persian ware. Furthermore, none can be reliably confirmed as belonging to the ship and in situ. Currently, all the pottery finds, including the Byzantine, and their relation to the ship, are suspect, but a thorough comparative study is in progress.

Figure 17.

Gaza ware and Yassi Ada pottery. (N. Sheizaf)

Figure 18.

Fragments of the mixed pottery remains. (N. Sheizaf)

Sequence of construction

After a careful examination of all the available planks, no evidence of any mortises or tenons, or any other edge joint, was found anywhere in the hull. It seems that the planks were all simply placed one next to the other and fixed to the frames by iron nails. The seams were later caulked. The total absence of edge-joints in the plank sections that were retrieved from the sea is being further confirmed in the conservation laboratory, though the work is still in its early stages. The presence of longitudinal reinforcements, such as keel, false-keel, a central longitudinal timber, two central stringers, two foot-wales, clamp, chine strake, and two wales, together with the method of joining the frames to the keel and the attachment of the planks and some ceiling strakes to the frames themselves (all by rectangular iron nails), suggest that the ship was built based on frames (see Lane, 1992: 8; although he refers to later Venetian galleys, the principle is similar). The original ship probably had more longitudinal reinforcements, but this analysis is only of the archaeological hull remains that were found.

The sequence of building may have been as follows: the keel and the endposts were joined together; the false keel might have been fitted at this stage, or at a later stage; the floor-timbers and half-frames were attached to the keel; planks were nailed to the frames from the outside, two nails in each plank for each frame, as in the butt scarfs. It is probable that some battens or strakes were attached to the frames in order to support their assembly and shape; for instance, the chine-strake may have been connected at an earlier stage of the planking. At least three half-logs, the chine-strake and two wales, were fitted from the outside in order to give the hull longitudinal integrity. The few additional frames that were not nailed to the keel were added at this stage; and maybe also the futtocks reinforcing the sides; the central longitudinal timber was laid on the frames and the endpost, and attached to the keel by iron nails, and the two central stringers were attached to the frames; ceiling strakes were laid on the frames, some fixed by iron nails; the foot-wales and clamp were attached on the inner side facing the chine-strake and wale 1, respectively; at this stage the mast-step sisters were positioned above the ceiling strakes. The mast-step itself was laid over the mast-step sisters, in the space left between the two central stringers. Finally, the seams were caulked, and the exposed timbers were coated with protective layers of resin.

Discussion

The dating of Dor 2001/1 is based on 14C tests (Table 2). Traditional 14C dating-methods conducted on random organic pieces gave dates over the range of AD 250–610. As the project progressed, AMS radiocarbon tests of short-lived organic material gave dates between AD 260 and 659. Averaging all samples using the OxCal method gives dates between AD 420 and 540 (pers. comm. Boaretto, 2005). Unfortunately, dendrochronology could not be used due to lack of wood remains with sufficient tree rings, and the absence of a database for this area in this period.

Despite the widespread distribution of ceramic sherds on the site, their connection to the ship is suspect. The lagoon is protected from the open sea by four small islands that do not prevent the action of the heavy winter storms. For example Byzantine amphoras (AD 324–638) have been found together with Persian-period ceramics (538–332 BC) in the same layer buried under 2 m of sand. Therefore these ceramic finds cannot be considered as in situ, and so cannot serve as indicators for accurate dating.

The commonly accepted theory regarding ‘transition in construction’ is based on the first wreck discovered that was built on pure skeleton, found in southern Turkey at Serçe Limanı, and dated to c.AD 1025. The frames had room and space of 33 cm, and were attached directly to the keel with iron nails. The hull was reinforced longitudinally by three stringers, two wales on either side, and a keelson. Pitch and caulking material were used in planking seams (Steffy, 1982a; Steffy, 1994: 85–91; Hocker, 2004a: 310, 313; Matthews and Steffy, 2004: 81–122; Steffy, 2004: 153–69).

However, earlier evidence of frame-based construction has come to light in recent years. The first wreck that raised the possibility of frame-based construction in the mid-1st millennium AD was Tantura A, dated to the end of the 5th to beginning of the 6th century. Its construction was based on frames, which were fastened with iron nails to the keel. No evidence for mortises or tenons was found; planks were nailed directly to the frames, and butt-scarfed at frame-stations, with caulking material in the seams. However, the change was explained by the fact that it was a small coaster which naturally adopted changes earlier than larger sea-going vessels (Kahanov, 2001; Kahanov et al., 2004). St Gervais B, dated to the beginning of the 7th century AD, was constructed ‘frame/skeleton-based’, with the frames attached to the keel by metal bolts. The few mortises that were found were widely spaced, and some tenons were unpegged. They had no pattern, and were quite insignificant in the construction, perhaps being employed for aligning some planks, or even in secondary use (Jézégou, 1985: 355; Jézégou, 1989: 142).

Yassi Ada of the 7th century AD is considered to be a significant landmark in the process, since it combined two methods of construction. The lowest strakes were aligned by widely-spaced unpegged mortise-and-tenon joints. The frames, attached to the keel, reinforced the structure, and were the basis on which the upper planks, which lacked mortise-and-tenon joints, were attached (Steffy, 1982b; van Doorninck, 1982; Steffy, 1994: 80–85). Tantura B, dated to the beginning of the 9th century, had a ‘frame-based’ hull. Its frames were nailed to the keel, and were further nailed from above through a keelson. The planks had butt-scarfs and L-shaped joints at frame-stations, and no edge-joints were found. Caulking was found in the seams (Wachsmann et al., 1997; Kahanov, 2000; Kahanov et al., 2004). The Bozburun wreck, dated to the 9th century, had no mortise-and-tenon joints, and its frames, with room and space of 30–40 cm, were attached to the keel by metal nails. Edge-joints in the form of polygonal treenails (‘coaks’) were found extensively in its plank edges. Thus it was based on frames, with skeleton orientation, but still had plank edge-joints (Harpster, 2002; Harpster, 2005).

The process itself went through several stages during the first millennium AD, which cannot be reduced to a simple linear chronological process. As more wrecks are being found, the more complicated it becomes, since many ships were built in combined methods (i.e. ‘strake-based’ with ‘frame-based’) (McGrail, 1997; Pomey, 2004). Scholars have suggested some ideas about the transition process; accordingly hulls should be studied from several aspects such as ‘principles of construction’ and ‘methods of construction’. Others suggest that they should be analysed according to their design, assembly-sequence and ‘structural philosophy’, that is to say how the shipwright planned to use the different hull-components for a good distribution of the working stresses (Pomey, 1988; Hocker, 2004b; Pomey, 2004; Pomey and Rieth, 2005: 155–86).

Using these analyses might make it easier to detect general principles which are shared by different ships built in different traditions, though by a similar process of construction. Table 3 shows a summary of relevant hull-construction details of seven wrecks, in which the gradual progress towards ‘frame-based’ construction can be seen. The data show that the transition in construction was well in progress by the mid-1st millennium AD. Characteristic features of the strake-oriented ‘shell-first’ hull, such as mortise-and-tenon joints, thick planks with long scarfs (‘S’, ‘Z’ and diagonal, connected by edge-joints), keel-garboard attachments, and frames not necessarily connected to the keel, are disappearing. Correspondingly, the frame-based construction characteristics of frames nailed to keel, thin planks nailed by small nails to frames, garboards not attached to keel, planking butt scarfs at frame-stations, and seam caulking, start to appear. However, the presence at the same time of both the Dor/Tantura wrecks, which adopted the transition, together with Yassi Ada and Bozburun, which still demonstrate plank-shell construction features, proves that the transition was not a linear chronological process.

Table 3. Summary of hull construction details
VesselDor 2001/1Tantura ASt. Gervais BYassi Ada 7Tantura BBozburunSerçe Limanı
  1. 1. Based on Reassembly Drawings I and II (Steffy 1982a, between pages 74 and 75 and personal comm.), and see Casson 1995: 216, based on unpublished information

Estimated date (AD)End of 5th- beginning of 6th C.End of 5th- beginning of 6th C.600–625625Beginning of 9th C.9th C.1025
Estimated length (m)161215–1820.518–2315–1615.66
Keel sided × moulded (mm)110 × 160110 × 180240 × 200220 × 355104 × 95180 × 290120 × 160
False keel width × thickness (mm)130–110 × 50–9042 thickness (?)
Keel and endpost rabbetOnly in endpostOnly in endpost+Only in keel edgeFlanged, T shapedOnly in endposts
Keel-garboard attachmentButtButtButtUnpegged mortise-and-tenon + iron nailsButt, nails in endsIron nails and treenailsButt, nails in ends
Keelson sided × moulded (mm)Central timber 150 × 180Central timber 240–310 × 400220 × 220120–202 × 157–180NA200 × 180
Wales width × thickness (mm)Two, 210 × 105 and 180 × 70Two, 110 × 90Four, half-logs, 200 mm diameterOne 100 × 85Two 145 × 131, 160 × 100
Frames sided × moulded (mm)Floor timbers 80 × 120 Half-frames 80 × 100 Futtocks 85 × 9090 × 95110 × 150–400140 × 14090 × 94120–170 × 140–220120 × 160
Frames room and space (mm)240324360300–3501260300–400330
Plank thickness (mm)25–302525–303525–3425–4040
Plank width (mm)100–17038–260200130–21040–360NA240
Planking scarfsButtButtNADiagonalButt, L-shapedDiagonal‘Z’ scarfs, butt
Mortise-and-tenonUnpegged, no system, widely spacedUnpegged, in lower section, widely spaced_Coaks_
Metal nails+++++++
Caulking in the seams++++++
Method of constructionBased on framesBased on framesSkeleton oriented, based on framesMixedBased on frames + strong keelsonMixedSkeleton

Dor at this period

The cargo of building stones might shed some new light on the Byzantine period in Dor, since the historical sources are rather sparse (Dahl, 1915). Until recently it was assumed that there was no settlement in Tel Dor throughout the Byzantine period (Stern, 2000: 323). However, excavations during the 1980s and ’90s suggest otherwise. According to the excavators, the Tel itself was indeed not populated at that time, but the lower city which lay to the east of the Tel was an urban settlement with public buildings, and a centre for a number of farms and villages forming its agricultural hinterland. It had a large episcopal basilica, whose earlier stages dated to the first half of the 4th century AD, serving pilgrims on their way to Jerusalem along the northern Mediterranean coast of Israel and the Via Maris. The basilica possessed the remains of two healing saints, a fragment of the stone of the Holy Golgotha, and was headed by a bishop (Dauphine, 1982–3; Dauphine and Gibson, 1994–5). The cargo suggests that building activity did take place at this time. Whether it was local to Byzantine Dora, or was part of the extensive building reflecting the prosperity of the age of Emperor Justinian in the 6th century AD throughout the Mediterranean (as for example, the blocks of marble found in the Marzamemi ship, off the coast of Sicily, Kapitän, 1969), is a question still to be answered.

Conclusions

This report covers four excavation seasons. Detailed studies of the finds and construction, and especially the section of the hull removed, are in progress. The conclusions should be read in this light, as an indication of work in progress. The authors intend to produce a final report which will deal with all the topics of this potentially important wreck in the fullest possible detail. Dor 2001/1 is dated to the end of the 5th to beginning of the 6th centuries CE. Shipping building-stones is an indication of the economy and society of the region, and especially about that of the Byzantine city of Dora, at the turn of the 5th century. There was quarrying and working of stones, and their transport by coasters to a construction enterprise, probably somewhere along the coast.

The hull remains of Dor 2001/1: construction sequence, attachment methods, scarfs and seam caulking; and the total absence of planking edge-joints or any other feature of plank-oriented ‘shell-first’ construction, lead to the conclusion that the ship was built based on frames. Thus it is the first wreck found in the Mediterranean, to date, that completed the transition from the earlier plank-oriented ‘shell-first’ to the later ‘frame-first’ construction.

Acknowledgments

This research was supported by Lord Jacobs, the Israel Science Foundation, the Hecht Foundation, the Sir Maurice Hatter Fellowship for Maritime Studies, and the University of Haifa, to whom we are grateful. The authors would like to thank the joint directors of the excavation, K. Raveh and C. Brandon. Thanks are due to John Tresman for reviewing the article and the English editing.

Ancillary