1Kibbutz Maʻagan Mikhael is located on the Mediterranean coast of Israel, about 35 km south of Haifa (fig. 1). It is known for the 400 BC Maʻagan Mikhael ship, discovered in 1985 (Linder 2003; Kahanov 2011). About 150 m south of this shipwreck site, another shipwreck was discovered by chance in 2005 by two local divers, Nathan Helfman and Yossi Batzir, who reported seeing framing timbers, ceramic shards and stones. In May 2015, this information was verified by a sub-bottom profiler survey conducted in collaboration with Drs Grøn and Boldreel of the University of Copenhagen, Denmark. The shipwreck was identified in 1.5 m of water, under 1.5 m of sand, about 70 m from the shoreline, and designated Maʻagan Mikhael B. In August 2015, a water-jetting survey exposed fragments of wood, rope and a pine cone (Cvikel et al. 2017).

2Wood and organic samples were dated preliminarily by ¹⁴C analyses to the 7th–8th centuries AD, which is the Late Byzantine – Early Islamic Period in the region. Eight samples were sent for dendrochronology to Prof Manning and Dr Lorentzen at Cornell University, and are currently being studied. Typological studies of the ceramics and amphorae have identified at least four types of Late Roman transport vessels, including LRA 1, 4, 5 and 13, dated to the 7th century.

3The main reason for excavating the Maʻagan Mikhael B shipwreck was its date: to evaluate the significance of the techno­logy of its construction in the period of transition in hull construction from shell-first to frame-based. Every shipwreck of this period can provide valuable new insights into the process of change in Mediterranean shipbuilding techniques, which took place during the second half of the first millennium AD. Until about 20 years ago, it was widely accepted that the first ship found of pure frame-based construction was the Serçe Limanı, dated to 1025 AD. However, the Byzantine and Islamic Period shipwrecks at Dor (Tantura) lagoon, have provided evidence of earlier frame-based ships (Pomey et al. 2012).

4A long-term programme of underwater excavation and research of the Maʻagan Mikhael B shipwreck was initiated. Four seasons of underwater excavations were conducted in May and December 2016, September 2017 and May 2018 by the Leon Recanati Institute for Maritime Studies of the University of Haifa (fig. 2). The site’s location was challenging, due to a small island, which creates a meeting point between the offshore (southwest) current and a local (northwest) current in all but calm weather. This fact, and inclement weather, allowed only 27 excavation days out of the 59 planned. In the course of the excavation seasons the well-preserved hull was partly exposed, documented, and some components retrieved for further study.

5The shipwreck remains, oriented roughly east/west (110°–290°), are 19.6 m long between extremities, and a maximum 4.9 m wide. The maximum depth reached was 3 m, and the keelson and keel have not yet been exposed. Among the hull remains in a good state of preservation are endposts, aprons, framing ­timbers, hull planks, stringers, the mast step assembly and bulkheads (fig. 3). A wide variety of finds were found: ceramic shards and complete amphorae, bricks, rigging elements and ropes, wooden artefacts, copper-lead alloy coins, food remains and animal bones. After documentation and careful consideration, some items were retrieved for further study.

1. Hull remains

1.1. Endposts

6The eastern endpost was composed of two timbers connected by an elaborate scarf, a variation of the “Jupiter” scarf. The upper timber was made of Aleppo or Turkish (Calabrian) pine (Pinus halepensis/brutia), and the lower timber was of walnut (Juglans regia). The endpost has been exposed for 186 cm. The maximum dimensions of these components are 12.5 cm sided and 23.5 cm moulded.

7The upper timber of the western endpost is missing. The lower timber, made of walnut, was partially accessible for docu­mentation. About 25 cm of its length was exposed. It is 13 cm sided and 12 cm moulded at the lowest exposed point, tapering to 10 cm at the top end.

1.2. Aprons

8Adjacent to each endpost were installed what seem to have been aprons, although their relation to the keel is as yet unknown. These timbers, made of walnut, were definitely used as internal reinforcements. About 40 cm of the eastern apron was exposed. Its dimensions are 24 cm sided and 22 cm moulded. The western apron was exposed for 49 cm, and it seems to be almost round, with a diameter of about 8 cm.

1.3. Framing timbers

9Well-preserved framing timbers were found on both sides of the hull, comprising at least 30 frame stations. They were made of walnut, except for one, which was of holly oak (Quercus ilex), with no clear indication of it being a repair. On average the framing timbers were 13.3 cm sided and 12.3 moulded, with an average room and space of 44 cm. About 1 m of the framing timbers was exposed, and it was seen that the moulded dimension increased towards the keel. The cross-sections of the timbers varied: some were made of partially worked thick branches; others were trapezoidal or rectangular. The faces that were connected to an adjacent frame or to a hull plank were carefully worked. On some of the framing timbers coating material and tool marks, perhaps of an adze, were visible.

1.4 Hull planks

10Sixteen hull planks made of fir (Abies sp.) were exposed. On average the hull planks were 17.6 cm wide and 3.2 cm thick. Many of the planks had scorch marks on their inner side, which might indicate “char-bending”. A single butt joint was detected with traces of two iron nails, one from each side: perhaps a repair. Apart from this, no planking edge joints were found.

1.5. Stringers

11Eight stringers made of fir were found fixed to the framing timbers by iron nails, as confirmed by XRF analysis (Fe 94.4 wt%). They were on average 18.3 cm wide and 10 cm thick. Scorch marks were visible on the worked faces, along with tool marks, which were especially apparent on the chamfered edges close to the eastern endpost. On the round face, facing the interior of the hull, were remnants of branches, most probably cut by an axe, and a thick layer of coating material. One stringer had Greek letters and symbols carved into it, as well as the word Allah in Arabic. The purpose of these marks is not known: perhaps they are marks of the shipwrights or crewmembers.

1.6. Mast step assembly

12The exposed section of the mast step assembly comprised the mast step itself, longitudinal support timbers and a lateral crutch (fig. 3). The mast step and support timbers were made of walnut, and the crutch of fir. The mast step was exposed for 3.5 m, and it continued further into the sand. It was 25 cm wide and 19 cm thick. Two mortises were cut into its upper surface: the smaller one was 27 × 8 cm, and 12 cm deep, probably housing a stanchion supporting a mast partner. The other mortise, measuring 37 × 10 cm and 6 cm deep, sloped downwards towards the west. This mortise probably housed the mast heel, and the slope suggests that the direction of the bow was towards the east. The crutch had a rectangular recess, about 32 × 9 cm, apparently to house the heel of a stanchion for the mast support.

1.7. Bulkheads

13Two transverse sets of vertical planks made of fir were found, most probably bulkheads. The eastern and western bulkheads were located 13.8 m and 5.5 m from the western apron respectively. The planks were typically 15 cm wide and 4 cm thick. The longest plank projected 54 cm above the sand.

2. Finds

14Various types of finds were discovered during the excavation seasons. They were documented on site, retrieved and registered, and are currently being studied.

2.1. Ceramic ware: shards and complete amphorae, and bricks

15Fifty-five amphorae were identified inside the ship (fig. 4a-b). This primary cargo was composed of six different types of amphora, including the types LRA 1, LRA 4, LRA 5, and LRA 13, as well as two types for which parallels have yet to be found. Thin section petrographic analysis suggests clay from Cyprus, Egypt and the Levant (Goren, personal communication, 2018). Some of the amphorae contained remains of their original contents, including raisins or grapes, olives, dates, ­pistachios, walnuts and pine nuts. In addition to the amphorae, a large assemblage of shards, juglets, bowls and cooking wares was found in the hull.

16Fourteen inscriptions have been found either painted on or incised in the pottery. The inscriptions are of Arabic and Greek letters and Christian symbols, implying a diverse trade network. Both Christian and Arabic symbols appear on one specific type of amphora, the LRA 13, probably produced in Cyprus (Demesticha 2005). The style of the Arabic inscriptions suggests a date in the Umayyad period (660–750 AD) (Abadi, personal communication, 2018).

17Two reddish-brown clay bricks (Munsell 5YR 4/4 and 2.5YR 5/6) were discovered near the eastern bulkhead. Both had charring marks, suggesting that they were part of the ship’s galley. Petrography of the rectangular brick revealed the origin of the clay to have been in Cyprus or the northern Levant.

2.2. Rigging elements and rope

18Two rigging elements, apparently turning blocks, were found amidships. They were 35 cm long and 10 cm wide. Both had a rectangular cross section with chamfered edges and showed evidence of tool marks. They comprised a single sheave and contained remains of rope. In addition, five well-preserved pieces of rope were found. All the ropes had three Z-twisted strands with an average diameter of 33 mm. The rope fibres were identified as hemp, coir and bast of unidentifiable species (Rast-Eicher 2016). Further study is needed to gain an understanding of the rigging of the ship.

2.3. Wooden artefacts

19Wooden artefacts were also found: 1) A roundel, 10 cm in diameter and 4 cm thick. Parallels to this were found in the Dor 2006 shipwreck and their use is still undetermined (Navri et al. 2013, p. 314); 2) Two worked wooden objects were found at the extremities of the hull, next to the aprons. The length of one of the objects is 20 cm, its maximum width 3.7 cm, tapering to 1.5 cm. Their use is still unknown.

2.4. Copper-lead alloy coins

20Eleven copper-lead alloy coins were discovered in the shipwreck and retrieved for study. Based on the portraits of Constantine the Great, the coins were dated to the mid-4th century AD. These coins were small change in circulation during the 4th–6th centuries AD. In some cases, they can also be found in contexts of the 7th century AD. Since coins are essentially mobile, the ship’s origin and dating cannot be determined using this information (Cohen et al. 2018).

2.5. Food remains and animal bones

21The food remains, which may hint at various aspects of daily life aboard a ship of this period, consisted of well-preserved olive pits, walnuts, grape seeds, peach stones, carob pods and pine cones (fig. 4b). Animal bone remains were also found. Among them were bones of goat, donkey, pigeon, chicken, sea turtle and tortoise. The bones most probably indicate food consumption. It is reasonable to suggest that at least some of it was part of the crew’s diet.

22The Mediterranean diet is based on food that is eaten according to the season. In view of the apparent remnants of fresh fruit found onboard, the ship seems to have sailed between September and December.

3. Suggested reconstruction of the original ship

23Preliminary dating, based on ¹⁴C analysis and the typology of the amphorae and ceramics, places the shipwreck in the 7th‑8th centuries AD. The dimensions of the Maʻagan Mikhael B’s components can be compared to other shipwrecks of the period (table 1). Considering the archaeological evidence, it may be concluded that Maʻagan Mikhael B was smaller than Pantano Longarini (31.5 m), but larger than Yassi Ada 1 (20.5 m). She was perhaps closer in dimensions to those of the Dor 2006. Based on the archaeological remains, it is suggested that the ship was originally about 25 m long, with a beam of about 7 m and draught about 3.5 m, and therefore capable of carrying about 150 tons of cargo.

24There is insufficient information as to the geometry of the cross section of the hull to suggest the construction tradition of the ship in the context of the transition from shell-first to frame-based. No planking edge joints have so far been found. As the study of the hull construction continues, the picture will become clearer and shed more light on the transition in ship construction.

4. Conclusions

25The ship was a 25 m long merchantman with a cargo of amphorae originating from various places, including Egypt and Cyprus. The circumstances of the wrecking event are yet to be determined. The Ma‘agan Mikhael B shipwreck promises to be an exceptional source of information regarding various aspects of ship construction, trading and economic activity, daily life, and seamanship and seafaring in the Mediterranean in the late 7th–early 8th centuries AD.

Acknowledgments

26The underwater excavations (IAA permits G-41/2016, G-40/2017 and G-26/2018) and research of the Maʻagan Mikhael B shipwreck are supported by the Israel Science Foundation (grant no. 1891/16), the Honor Frost Foundation, a Dov Shafir Fellowship, a Natan Rotenstreich Scholarship, a Sir Maurice Hatter Fellowship, the Research Authority of the University of Haifa, and anonymous donors, to whom the authors are grateful. The authors would like to thank B. Lorentzen, S. Manning, R. Yeshurun, I. Abadi and Y. Goren for their valuable assistance, and J.B. Tresman for the English editing.