THE UNDERWATER EXCAVATIONS IN CAESAREA 1995-1996
Analysis of data gathered in previous years (see CMS News, report no. 21, 1994:4) has raised many new questions which we have attempted to answer in the past two field seasons of underwater excavations at the ancient harbors of Caesarea. This has led us to focus on a series of new issues.
Area K, at the northern tip of the main western mole of Sebastos is the site of a huge mass of collapsed building stones. While studying the series of 'single mission' wooden barges that were used as caissons for this pozzolana built artificial island, we were intrigued by the fact that most of these caissons were displaced, tilted and segmented. This evidently occurred long ago, judging by the metal remains of a late 1st century CE merchantman, which foundered at the site.
Wooden caisson at K-3. (Photo: S. Ovadiah)
During the 1995 season we continued excavating at the gap between K-3 and K-5, all the way down to beneath the floors of the two neighboring forms. Scores of displaced wooden boards from the southern sheer of K-3 were exposed, in a stratigraphic context of sherds and broken vessels dated to the early 2nd century CE. A deeper probe was made under the wooden floor of K-3 caisson which exposed a series of alternating laminas of seashell particles and fine sand. These repetitive, single phase depositions were laid by the surge in a hollow created by undertrenching currents, following the displacement and tiltage of K-3 towards the NE. Below these laminas, at least 1 m deeper than the caisson floor, is a coarse deposit of shingles and small pebbles, mixed with bones and wave eroded sherds from the late 1st and early 2nd century CE. The combined structural and sedimentological data found at the site, enable us to suggest a hypothetical sequence of displacement and decomposition of the wooden forms at area K, with an estimated date in the 2nd century CE for its final phase.
During the fall of 1995, a ten day underwater field session took place. The wooden caisson at the NW tip of the northern mole (area G), first studied by CAHEP in the early eighties, was reinvestigated. The sleeper beams along the mid-section of the western side were exposed and carefully studied. Though clear evidence was found to suggest some violent damage inflicted on the wooden structure at that part, no clearly dated context was disclosed. We do have, however, a general notion that it took place relatively early, during the Roman era. Another probe was made around the SE corner of the rectangular block. The same type of stratigraphic evidence for undertrenching was exposed along and under the eastern edge of the block, with scores of early 2nd century CE pottery and wave eroded earlier pieces, deposited well below its floor.
This new information correlates nicely with data gathered
during earlier seasons. It suggests a significant upheaval some time around
100 CE or just prior to that date, which resulted in major damage and considerable
displacement (and subsidence?) of the main architectural components of
the entire western half of Sebastos and perhaps beyond.
Hypothetical sequence of decomposition and undertrenching at
area K. (Drawing: C. Brandon)
Area U is a massive spill of large blocks, dissected cement units and building stones, creating an 80 x 35 m rampart of a roughly rectangular shape, that rises some 4-5 m above the surrounding seafloor, half-way along the external (western) perimeter of the main mole of Sebastos. During the 1994 and 1995 seasons some twenty probes were carried out along the sides of the spill, in order to determine whether this was also a wooden, caisson-based, artificial island, established as a constructive component of the enclosed harbor basin.
Though in many of the probes cement blocks were exposed and found to be of the same 'three layer' type as in area K, indicating their origin within wooden caissons and 'single mission' barges, no wooden forms were traced anywhere. Yet, in many cases plaster coated, cut kurkar blocks were found buried in the sand at a depth of more than 10 m below the waves. These are quite clearly components of upper structures, probably the warehouses that were built along the main mole of Sebastos, according to Josephus. Their present exceptionally deep position indicates that the seafloor was much lower and the vertical displacement at that area greater, than elsewhere in the harbor. This fact correlates nicely with the 1976 geological survey that suggested a secondary fault line along the very western side of Sebastos, parallel to the main line which dissects the eastern side of the main harbor basin. The pottery found in this context is mostly of a relatively early date, including a complete Rhodian amphora with both handles, stamped with an eponym from the Lelegian city of Myndus in Caria.
Over the past two seasons a rather tedious search for the remains of wrecked vessels has been carried out along the seabed within the subsided main harbor basin of Sebastos. We aimed to find vessels that were lost either when the harbor was still intact, or foundered later when attempting to sail over the already subsided moles. An extensive series of water-jet probes were carried out within the original harbor basin, in a search for buried remains of ballast stones or sea borne cargoes.
The hidden, potential 'targets' were marked, plotted and
eventually probed using dredging units within cylindrical metal caissons.
Each probe was dug while carefully considering stratigraphic factors. Ample
samples of sediments were extracted from every layer in each probe. These
samples are currently being analyzed for grain size and microfauna, in
order to determine and reconstruct the paleoenvironment. The surveyed area
has been subdivided into areas RS, RN and RW. Of the various probes which
have reached the level of the preharbor seafloor, only one, RN-2, has yielded
significant remains of what seems to be the wreckage of a 6th century merchantman.
Basalt ballast stones, some broken jars and scores of iron nails are all
that has been retrieved from that site to date.
Probing in area RN, within a metal caisson. (Photo: J.J. Gotlieb)
Most of the underwater work of the 1995-1996 season concentrated on the intermediate harbor basin and the line of division between it and the inner basin. This area of rather shallow water correlates to the present day anchorage, which is due to be renovated in the near future. There are plans to reopen part or all of the presently landlocked and over-built inner basin.
Six sites in the western part of the intermediate basin, all to the north and NE of the present quay, were probed in order to study the stratigraphy of accumulated harbor debris on the seafloor and the topography of the bedrock below. The area, designated QN, is covered with scattered building stones of various sizes and large chunks of kurkar, which seem to date from the Crusader era on. No sherds, or other artifacts were found, nor fine-grain sediments which usually characterize the bottom of harbor basins. It is as if this area had been thoroughly dredged, either by the later people of Caesarea (during the Crusader period?), or flushed by the rip current. Another possible explanation for this stratigraphic lacuna, is the alleged dredging of this anchorage during the Early Islamic era, sometime before the late 9th century CE.
The topography of the bedrock suggests a fairly even elevation,
presently 4.6 - 5.1 m below MSL, with something akin to an underwater gorge,
sloping toward the SW and reaching a maximum depth of almost 7 m at the
base of the Herodian Quay, at area Q.
Area T is the general area of the eastern part of the present anchorage, adjacent to the bathing beach. The main man-made feature apparent in this area is a wide seawall. It is incorporated into the wall of the Crusader city at the seashore on the north side and crosses the bay almost to its southern end and the adjacent collapsed remains of a currently submerged, rectangular tower.
Two parallel trenches (TS) were excavated to the east of the tower, crossing the northern edge of the alleged navigation channel, which once led into the inner basin. An attempt was made to trace any structural remains of a retaining wall. Extensive fill, composed of building materials and scores of domestic pottery, dated to the last phase of the Byzantine era, was exposed, but no in situ structural remains were found. The bedrock of crumbled, ill-cemented eolianite kurkar, which is at an even elevation of about 2.5 m all over the bottom of the inner basin to the lee of the Medieval seawall, was found to slope down towards the south, to an additional meter in depth. This may have been the alleged channel, quarried in order to enable better navigation and perhaps also to improve the flow of the outgoing flushing current from the inner basin.
TN is the marking for another trench, begun in 1993 and excavated perpendicular to the northern part of the beach in a westerly direction, towards the Medieval seawall. During the 1996 season, this trench was extended further to the west in order to expose both sides of the seawall and study the accumulation of debris under and adjacent to it.
TN-1 is the demarkation of the westward trench (TN). On the lee side of the seawall the fill is composed of four layers: a. Surface material of mixed, circulated and marine coated building stones; recent deposits and well eroded sherds - at 1.4 - 1.9 m below MSL. b. A fill comprised of ashlar blocks from terrestrial buildings, some coated with plaster; mixed with sherds of the last Byzantine phase (6th - 7th centuries CE) - 1.8 - 2.3 m below MSL. c. A fairly compact layer of mud, mixed with fine sand and Late Byzantine sherds of brackish or terrestrial origin - 2.2 - 2.7 m below MSL. d. A rather thin layer of decomposed kurkar particles and some large limestone pebbles, mixed with easily dissolved lime and a few Early Roman sherds; this layer integrates with the crumbled bedrock of ill-consolidated, laminated kurkar.
The basic course of ashlars, of which the eastern face of the seawall is comprised, lies in the second layer from the top, at a depth of 2.2 m below the MSL.
The seawall itself was over 6.5 m wide when built, probably with an east facing rampart at its lee (included in that width). The wall was based on reused columns laid parallel along its weather side. The columns, most having been displaced, were found just below the present seafloor, on the western side of the wall. Several wooden planks were exposed, running beneath both the columns and the wall. The same stratigraphic sequence was exposed under and to the west of the eastern side of the wall. Yet, on the western side far fewer terrestrial filling materials were evident. A much thicker mud layer was apparent (reaching a depth of 3.6 m below MSL), with the bedrock almost 4 m below the waves.
TN-2 designates an additional probe to the west of the columns, at the western face of the seawall, some 20 m north of TN-1. In this trench, excavated during the last phase of the 1996 season, we were surprised to find an extensive layer of wooden boards covering the entire 3 x 6 m excavated area. This layer seems to continue eastward, under the base of the seawall. The topmost boards lie 2.3 m beneath the water and are covered by some kind of cloth. There are at least two additional layers of wood underneath, some of which terminate to the west in a scarfed joint. No tree nails, frames or other fastening features were observed. If these boards are remnants of a ship's hull, it is quite probable that a 'skeleton first' technique was employed in its construction. This became standard only during and after the Byzantine era. The boards were covered by loose sand, building stones and well eroded sherds. Quantities of Late Byzantine pottery, including some intact vessels were found within this stratigraphic context. Among other small finds are several well preserved glass vessels and a 'lump' of over seventy Byzantine bronze coins, most of which are of the nummia type, minted up to 608 CE. An ambiguous find, is a gold solidus of Constans II, dated to 651-654 CE. This coin, minted at least twelve years after Caesarea fell into Arab hands, might help to confirm our former hypothesis that the extensive debris found silting the sea basin at area T, was actually placed there deliberately by the Arab settlers of Caesarea. This would have been done in an attempt to block the local anchorage and impede a Byzantine, seaborne raid on the city. Whether the attempt was made before or after an actual raid took place in 685 CE, is still an open issue. Some later Arab and Crusader sherds found at higher levels might illustrate the continuation of seaborne activities at this site. Perhaps this was permitted by partial dredging of the basin, sometime in the 9th century. The dredged fill would have been dumped on shore just south of the fortified city.
A gold solidus of Constans II (651-4), from area TN. (Photo: J.J. Gotlieb)
SW is a single probe, made at a randomly chosen location, some 50 m to the west of TN-2. The aim of the probe is to study the stratigraphic sequence of this section of the intermediate basin. At 2.3 m depth, the sandy seafloor is littered with scattered building stones and rubble. Beneath only 0.3 m of this loose, wave-circulated mixture, is a distinct layer of mud, in which is embedded a large quantity of broken Byzantine jars and pottery. This context, at a water depth of 2.6 - 3.3 m, includes some earlier clay vessels. These are mostly imported amphoras and local jars, dating from the 4th and 5th centuries CE.
Beneath the mud layer is a sand layer over 1 m in thickness. This loose sand was a continual problem, flowing into the dig and making the probing work rather syziphian. Yet, towards the end of the 1996 season, we did succeed in reaching an extensive mass of Gabro type, black basalt blocks of non-local origin. These stones are quite probably the remains of ballast either jettisoned from a vessel or sunk within the hull of a ship.
Areas RN, TN-2 and SW will be further investigated
An intact Byzantine jug within wooden remains in area TN-2, from the east. (Photo: A. Raban)