A dive conducted prior to the survey, with the participation of the divers who discovered the ancient artifacts, revealed that recent storms had displaced sand from an area measuring 50 × 70 m at a depth of 7–8 m, on the north side of the secondary (northern) breakwater, close to the juncture of the breakwater with the natural kurkar reefs (Fig. 2).
The salvage survey recovered numerous artifacts that apparently originated in the large cargo of a Roman ship. Two groups of artifacts were identified. The first group comprises objects used to operate and maintain the ship, including anchors, sounding weights, a lead brazier, fasteners and tools, lead sheathing for lining the hull and bilge pump components. The second group comprises a large variety of objects from the ship’s cargo, including bronze statues, bronze lamps, coins, raw glass, lead mirror frames, fishing gear, stone objects and a steelyard. Most of the objects were well preserved due to the thick sand cover (3 m and more) that protected them from the sea and human activity.
The extension of the survey during the same year employed new technologies for documentation and mapping, revealing that the seabed at the site is sandy and particularly thick. Data collected from the meeting point of the breakwater’s base and the reefs show that the site is a sand trap, and only extreme sea conditions cause a considerable sand shift. Water depth measurements carried out over several years showed that in 2010 the depth at the site was 2–3 m (Steinberg 2010:56–59), at the end of 2014 it was 4–5 m and at the time of discovery of the site in 2016 it was 7.5–8.0 m. Thus, over these years, a 5–6 m displacement of sand occurred, uncovering the content of the ship.
Objects Used to Operate and Maintain the Ship
Anchors. Three iron anchors were found, two complete and one broken. The largest anchor (height 2.9 m; Fig. 3) was found at the center of the site, at a depth of 7.5 m; it was used to anchor the ship. Its arms (length c. 0.9 m each, span 1.45 m) end in sharp flukes. The stock (length 1.22 m) is connected to the shank and half of it is missing, indicating that its full length was at least 2.5 m; stock length is usually equal to the anchor’s height. The shank has a square cross section (width 8 cm, thickness 6–7 cm). An iron ring (diam. 0.3 m) at its head is for tying a rope. The second largest anchor (height 1.4 m, arm length c. 0.5 m, span 0.8 m; Fig. 4; Kapitän 1984:43, Fig. 8b) was found attached to a rock in the northeastern part of the site. An iron ring at its head (diam. 0.12 m) was for tying a rope. The third anchor was found broken nearby. All three anchors date from the Roman imperial period and are of Types B and C according to the accepted typology (Kapitän 1984:43, Fig. 8). Ten additional stocks found (length c. 2 m) indicate that the ship carried additional anchors.
Sounding Weights (Fig. 5:1–3). Eight sounding weights were found; they were used for measuring water depth and determining the seabed composition with the aid of a sticky substance (fat or wax) that was inserted into the hollow on the bottom of the weight. Sediments attached to the sticky substance allowed to determine the character of the seabed. Information on the topography of the seabed was crucial to mariners when navigating close to shore, as warning against shallow water, reefs or sand (Oleson 2008; Galili, Oleson and Rosen 2010).
The eight sounding weights are made of lead and are bell-shaped, with a rounded, triangular or flat top. Some have a perforation for attaching a rope, others a bronze ring. Weight No. 1 is the largest and heaviest (24.4 kg), with a hollow at the bottom for inserting a sticky substance and holes for inserting nails to secure this substance. Weight No. 2 is the smallest (2.72 kg), with a hollow bottom divided by two crossed partitions that bear five protrusions. Weight no. 3 (9.9 kg) is round, with a narrow, round hollow to hold the sticky substance.
Lead Brazier (Fig. 6). The well-preserved rectangular cooking brazier still had soot and ashes in it, evidence of its use. It has double walls that form a closed vessel with a cylindrical ‘chimney’ into which water was poured. The water functioned as coolant, preventing the melting of the lead (melting point 327.5° C). The outer shell is a rectangular compartment for placing wood or charcoal for burning. An experiment conducted with a model brazier with similar technical characteristics showed that with charcoal as fuel, the maximum temperature reached was 104° C, while with wood it did not reach above 101° C (Mosyak et al. 2017). In either case, when operating the brazier with water in the inner space, the temperature did not become high enough to melt the lead and cause a fire or damage the wooden hull. Numerous such braziers were found during underwater surveys along the Israeli coast—nine on the Carmel coast, three at Ashqelon, two at Yavne-Yam and two in Gaza—but none were found in an archaeological context (Galili and Sharvit 1998; 1999a; 1999b; Rosen and Galili 2007; Galili and Rosen 2011; 2012; 2015). Only three braziers were found elsewhere: two in Turkey and one in France (Leonard 1973; Joncheray and Joncheray 2004). This is the first discovery of a brazier within an entire ship assemblage, allowing it to be dated.
The lead sheet on which the brazier was placed (0.32 × 0.46 m, thickness 0.55 cm) bears the marks of the brazier’s bottom, as well as soot indicating its operation. The sheet is particularly thick, to prevent direct contact with the ship’s wooden hull; it is the first of its kind to be found.
Fasteners and Tools. The numerous fasteners found include hundreds of copper-alloy nails with a square cross section and of varying length. Such nails are characteristic of Roman ship construction and were used for joining different timbers of the ship’s hull. The smallest nails (length 12 mm; Fig. 7:5) resemble tacks that fastened the lead sheathing that covered the outer surface of the hull; similar tacks were found in a Roman shipwreck off the Carmel coast (Galili, Rosen and Sharvit 2010:62–67). The larger nails, used for fastening larger components, are round with a square head (not illustrated). The way the nails are bent signifies their use: some are clenched (bent 90°), and some are double clenched back into the timber (Fig. 7:6); one is a one-way nail, preventing its extraction (Fig. 7:2). Some nails have a flat, square head that allowed hammering the nail head into the wood (Fig. 7:4). Some nails (Fig. 7:1, 3, 7) were found without signs of use and were probably part of the maintenance kit of the ship's carpenter. Additional finds include an iron hammer with a nail extraction claw, wide and narrow iron chisels and a wooden handle, which were used, inter alia, for caulking. Since iron undergoes extreme oxidation in a marine environment, all the iron tools were found embedded in a thick conglomerate of sand and shells that were cemented by the oxidation products; in some cases, all that was left was the tool’s outline in the conglomerate.
Lead Sheathing. Among the finds were scores of lead sheets (thickness 1–2 mm; Fig. 8) and lead rolls of various sizes. Lead sheathing is known mostly in ships dated to the third–second centuries BCE, in which two-thirds of the hull was lined with lead sheets (Kahanov 1999; Kahanov and Ashkenazi 2011). The lead sheets found at the site sank to the seabed after they were torn off the hull when the ship sank and disintegrated. Some sheets had perforations left by nails that affixed them to the hull, and some bore herringbone patterns on the inner side (Fig. 8:2). Rectangular and rolled lead sheets were used for patching the hull sheathing and repairing other lead tools or installations on the ship. Sheathing ship hulls with thin lead sheets was practiced from the sixth century BCE until the third century CE. The sheathing covered the lower part of the hull’s exterior, from the keel to slightly above the waterline. The poisonous lead kept marine organisms from attaching to the ship and damaging the wooden hull, similarly to modern antifouling paint (Kahanov 1993:244–247).
The Bilge Pump Assembly. Parts of the pump assembly, including lead tubing (total length c. 10 m) and water containers, were found at the center of the site. The containers were made of lead sheets. The lead tubes, which were attached to the containers, were formed by rolling lead sheets (width 0.21 m, thickness c. 0.5 cm) around a rod (diam. c. 5 cm), leaving an overlap (width c. 3 cm) and soldering the joint to form the final shape and seal the tube. Some tubes bore solder remains where they were attached to other tubes to form longer units. The longest piece of tubing found is 2.44 m long (Fig. 9). Lead tubes of a similar diameter, made of thinner sheets (thickness c. 2 mm), were also found.
The Ship’s Cargo
Bronze Statues. Hundreds of bronze statue fragments found include the right hand of a female figure (Fig. 10:1); a figurine of Luna, the moon goddess (Fig. 10:2); the left hand of a male figure (Fig. 10:3); a hollow figurine, probably of Dionysus, which decorated a vessel with the upper part of the head serving as the lid (Fig. 10:4); a fragment of a statue’s face, possibly that of Augustus (Fig. 10:5; Barnett 1996:15); and part of the hand of a youth or boy holding a dove (Fig. 10:6). The body parts comprise at least five statues, including four male and one female. A comparison to average human dimensions indicates that the statues exceeded life-size proportions (c. 120%). The statues were apparently broken into small pieces in order to accommodate them in baskets or barrels for ship transport (Robinson and Wilson 2011:3). Lead-alloy lumps were found attached to some of the statue pieces, indicating that the statues were meant for recycling.
Bronze Oil Lamps (Fig. 11:1–3). Three large bronze oil lamps were found. Two of them (Fig. 11:1, 2) are similar in size and shape, except for the decorative figurine. Lamps 1 and 2 have high ring bases, two arms ending in projections resembling lion claws and nozzles shaped like wide arcs. The handle of each lamp is surmounted by a figurine attached obliquely, as if looking at the nozzle of the lamp. The handle of Lamp 1 is shaped as a stem and flower, from which emerges a figurine of Sol, the sun god. The handle of Lamp 2 is decorated by a hitherto unidentified deity, draped and diademed. Lamp No. 3 is shaped as the head of an African man. Judging by their artistic style, the lamps date from the first and second centuries CE.
Coins. Two lumps of coins contain hundreds of coins (c. 15 kg; Fig. 12) and preserve the shape of the ceramic vessel in which they were stored. Thousands of additional coins were found scattered on the seabed around the lumps. It appears that the vessels containing the coins broke when the ship sank; the coins in the upper part of the vessels scattered and those at the bottom concreted into a solid mass. Of the assemblage, 992 coins were identified (see appendix), most dating to the first quarter of the fourth century CE.
Raw Glass. The 26 chunks of raw glass found (total weight c. 150 kg) are mostly green, yellow and bluish (Fig. 13). The largest weighs 14.62 kg. A large chunk with a dark, almost black hue originates in northern Sinai. Raw glass chunks have been found all along the Israeli coast. However, none were found in an archaeological context, therefore dating them is impossible. The present assemblage is the largest found to date; it is the only one that is part of a dated cargo and related to a specific commercial port that apparently served as a loading station for glass trade. The glass chunks are evidence of large-scale trade with raw-glass-producing centers and overseas markets (Galili, Gorin-Rosen and Rosen 2015). Most raw-glass production in the region took place between the Sharon plain and the ‘Akko coast. During the Roman period, the Levantine coasts of Egypt, Palestine and Syria were the main suppliers of raw glass. Glass ingots were loaded onto ships that departed from harbors and anchorages close to the production centers toward the manufacturing centers around the Mediterranean basin (Radic 2012:17–22).
Lead Mirror Frames (Fig. 14). The 58 round and eight square frames form the largest assemblage ever found. The round frames (outer diam. c. 6.3 cm) are flat and thin, their back is smooth and their front is decorated with vegetal motifs; they seem to have been manufactured in the same workshop and perhaps even cast in the same mold. The square frames slightly differ from one another. Most have a mask or a face decoration in the corner and one is decorated with flowers. They are rare and only a handful of similar frames were found and published (Vinokurov and Treister 2015). One of the round frames was found with a fragment of a round glass mirror.
Such mirror frames have not been previously found in Israel. They are primarily found in northern Italy, as well as in Rome, Ostia and along the Danube, and have been discovered as far as the Crimean Peninsula, Cyprus and Syria. Their production center appears to have been in Rome and its environs (Buora 2021:189, Fig. IV.1–3).
Fishing Gear. Fishing gear seems to have been an important and significant part of the equipment loaded on ancient merchant ships, as it helped supplement the fresh food available during the journey. Like other similar assemblages, the Caesarea ship yielded fish hooks of various sizes, lead sinkers, dozens of lead weights for fishing nets (Fig. 15:1–13) and bronze knitting needles (Fig. 15:14) for mending fishing nets. Two hooks are extremely large and only a few like them were previously found. One hook (length 12 cm) was attached to a chain (length 0.6 m; Fig. 15:15) with a swivel joint in the middle, preventing the chain from twisting and breaking. This large hook was intended for fishing large, deep-water fish. It is the only hook of this kind found in Israel so far; there is only one other identical hook known from the Mediterranean basin, found in excavations south of Pompeii (Stefani 1990:14). The dozens of cylindrical and folded rectangular lead weights, some decorated and some smooth, include a group of identical weights which were likely attached to a fishing net that was not preserved.
Stone objects. The stone objects comprise a white marble colonnette with a base (length 1.4 m, diam. c. 0.12 m; Fig. 16); its capital (height c. 0.2 m; Fig. 17) was found broken next to it. The capital is of the Corinthian order. The centers of two of its sides are decorated with vegetal motifs, while the two other sides show Serapis and his consort, Isis. The top of the capital bears burning marks, indicating that a vessel was placed on it. It seems that this was part of a stone altar that included a table and two colonnettes on which the bronze lamps (above) were placed—possibly part of a ritual corner on the ship, or part of the cargo. Such altars were common, and many appear on Roman-period coins.
Steelyard (Fig. 18). The steelyard comprises a rod with a square cross section, divided into two unequal parts. Weighed items were placed in a pan hung from the short arm by three chains and a hook. An additional hook was used for hanging the steelyard, and another hook served for hanging heavy items. A weight was hung on the long arm and could be moved along the arm. Items found with the steelyard include a long and narrow rectangular weight that ends in a figurine, made of bronze and filled with lead, which could be calibrated; a similar weight was found on a Late-Roman shipwreck off the Carmel coast (Galili, Rosen and Sharvit 2010: Fig. 50). Additional finds include a bell-shaped lead weight with a bronze hanging ring (778.41 grams), a bronze strigil, various keys, rings and iron tools.
All the artifacts were found within a small area, where the ship sank to seabed. Based on the size and weight of the anchors and numerous hull components, such as the bilge pump and various fasteners and nails, it appears to have been a medium-sized merchant ship (length 20–30 m, 75–200 tons displacement) from the Roman period, of a type depicted on mosaic floors, such as the Lod mosaic (Friedman 2004).
As in other Roman cities, the streets and squares of Caesarea were embellished with many statues of marble and bronze. Such statues would show emperors and high-ranking officials (Gersht 1996; 2008; 2015). Numerous excavations conducted at Caesarea over the years have uncovered marble statues, some of which were damaged by sawing for reuse in construction or in limekilns. Life-size bronze statues were few (Gersht 2011:77­–79). The discovery of hundreds of fragments belonging to at least five life-size bronze statues, as well as lead-alloy lumps found on some statue fragments and numerous other lumps, indicate that the ship’s cargo included many objects sent for recycling.
The large coin hoard date the merchant ship to the beginning of the fourth century CE. The disparity between the accepted date for the use of lead sheathing—the second century CE at the very latest—and the dating of the ship’s sinking to the fourth century CE seems to indicate that the use of lead lining of hulls continued up to this date. It is reasonable to assume that the lead sheets found at the shipwreck were not meant for recycling. However, the disparity between the date of many of the finds in the first and second centuries CE and the date of the coins apparently indicates that the larger part of the bronze pieces were looted and collected by local scrap dealers and sold for recycling.
The raw glass chunks found at the site were traded and sent for the manufacture of glass vessels. The seabed along the Israeli coast yielded numerous raw glass chunks, but none in a dated assemblage of a shipwreck. This assemblage is the largest and the only one found so far in an archaeological context and in the context of a commercial port that was a loading station for glass trade. Based on the distribution pattern of the square lead mirror frames, which is primarily concentrated in northern Italy and Rome, and on the retrieved objects related to deep-sea fishing of large species that live in the western Mediterranean, it is plausible that the ship began its voyage in one of the ports in western Italy, sailed to the eastern Mediterranean and anchored at Caesarea for trading purposes.
The ship seems to have met emergency conditions when exiting the harbor and was carried east to the secondary breakwater and the reefs. As it was swept by the waves, the crew dropped the main anchor, which was found broken. Since wave energy at this spot is considerable, the ship sank with its cargo. Many of the wooden parts of the hull were likely abraded and consumed on the seabed and near the shore, and some may have been collected by locals for reuse. A medium-sized merchant ship could transport dozens of tons of cargo. From this ship, only about 1.0–1.5 tons of cargo were recovered so far. Thus, a considerable portion of the ship and its cargo may still be lying on the seabed, under a cover of sand.