Herod's Harbour (Sebastos)

The harbour front reached from the park's northern gate is a working quay with restaurants, galleries and an artists' court along it, run by the Caesarea Development Corporation rather than by the parks authority. Herod's outer basin lay west of it, and the breakwaters that enclosed it are shoals now rather than structures.

The middle of Herod's three basins was based on a natural bay and sits under the modern anchorage. The Old Caesarea Diving Club, which runs the underwater trail here, notes that the original Herodian paving of the great breakwater can still be seen beneath the modern concrete quay, and that a round tower about 13 metres across still stands at the mouth of the inner anchorage, probably on an earlier fortification of Straton's Tower.

The innermost basin was never enclosed at all: it was dug, and it is now almost entirely dry land, under the seafront buildings and the lawn. At its landward edge, at the foot of the temple platform, a mooring quay survives with a tie stone in it, and under the tie stone a horizontal notch worn by standing water. That notch is a measurement. It records where the sea sat when ships were tied here, which is why what to look for on land is a water line rather than a building.

The ROMACONS team describes the setting bluntly: a shifting sand beach with no mitigating physical features, exposed to the longest fetch in the Mediterranean and scoured by a strong current carrying sand from south to north. Their reading is that the site was chosen for political reasons and not because nature favoured a port there. Josephus puts the same problem from the sailor's end: between Joppa and Dora there was no haven fit for ships, and merchants were generally forced to ride at their anchors in open sea.

There was a commercial case as well. The diving club puts the motive in trade: a principal port inside Herod's kingdom for luxury goods moving from the Indian Ocean to the Mediterranean, the spices, costly textiles and gemstones whose demand had risen across the empire since Augustus.

When it was finished is genuinely unsettled. Josephus gives twelve years for the whole programme, and the diving club dates harbour and city to 22 to 10 BCE on that basis. ROMACONS puts the harbour at roughly 23 to 15 BCE, Roman Ports has it begun in 22 and completed in 15, and the Caesarea Development Corporation says ten years of engineering. None of the accounts hides the disagreement.

Roman hydraulic concrete was lime, water and pozzolana, a volcanic ash rich in aluminosilicates, with rubble added as aggregate. It sets under water, which is what made a harbour on this shore thinkable. Getting it into place took wooden formwork, and Caesarea is the site where at least three systems have been identified in one harbour. ROMACONS calls the reason unclear, and the variety itself one of the project's most striking features.

In October 2005 the study took five cores from submerged blocks here, three from the southern breakwater and one from the block at the terminus of the northern one. It was its first campaign in which every sample came from well below sea level, recovered with scuba. The results were unflattering: all five tested substantially weaker than the Italian cores in the project's database, and two showed less than half the compressive strength of its benchmark sample from Santa Liberata.

The field notes are more interesting than the verdict. Large lumps of lime in all five cores carried thin alternating black and white stripes the team had not seen at any other site, apparently worked into the mix several times and then left unreduced. Small unmixed lime lumps were commoner here than in Italy, and so were numerous uniform small voids. The explanations are offered as possibilities rather than findings: hurried mixing on an exposed site, mixing done on board ship, a workforce unused to the material, or warm water setting the mortar before it could compact.

Divers meet the consequence directly. At the northern end of the main breakwater there are voids inside the blocks that divers call caves, which the diving club attributes to weathering of a second pour of poorer quality than the first. The aggregate was local kurkar, and the mortar appears not to grip it as well as it grips Italian tuff. The ROMACONS summary is worth keeping: an inferior concrete, and still a remarkably durable one.

• Planks driven into the seabed and braced into a box, then filled: the method Vitruvius describes
• A double-walled hollow box built on shore, towed out, and sunk by filling the cavity with mortar
• Barges planked with mortise-and-tenon carpentry, part-filled in the shallows and towed into position

The outer basin was made by two breakwaters. The main one starts in the south-east and swings in a broad arc to the north-west, carrying a sea wall with towers on its seaward face and, on its inner face, a loading quay with warehouses and a promenade. At its southern end, where the citadel and the Time Tower now stand, was the largest tower of all, the Drusion, which the diving club suggests may have been the lighthouse. Estimates of the entrance channel's width run from 30 to 80 metres, and two towers stood outside the mouth to mark the approach.

The northern breakwater is shorter and straighter, and probably carried the harbour administration building. Niches cut into its stones may have held pulleys, and the pulleys an iron chain that could be drawn across the entrance. The diving club offers that as a suggestion, not a settled fact.

Two pieces of engineering deserve more attention than they get. The first is the Procumatia, a low forward breakwater about 6 metres wide that reached only to the water surface, set roughly 10 metres south of the main mole and parallel to it. It broke the spray that would otherwise have interrupted loading, and it could be repaired each season without touching the main structure. Josephus names it.

The second is the answer to silting, the problem that kills harbours on a sand coast. Cut into the rock at the south-eastern edge of the middle basin is a channel whose sill stands clear of the seabed and whose head reaches 20 to 30 centimetres above sea level, with vertical slots for wooden doors to regulate the flow. Channels like it let clean water in and set up a constant outward current that carried sediment away through the mouth. The evidence that it worked is a deposit a metre thick at the harbour entrance, full of silt, pottery, metal and pieces of ships swept out of the basins while the port was still running.

The harbour started going down immediately after it was built, and the diving club lists the candidate causes without picking one: a geological fault about 150 metres off the present shore, a series of earthquakes, possibly tsunami waves, an unstable sandy seabed, and the sheer weight of the structures. Add the weakness the cores show. Researchers still differ over how much to give each, and the site's operators do not pretend otherwise.

The timing is better established than the cause. By the end of the first century CE the breakwaters were already submerged, which is known from the cargoes of ships wrecked on them: near the mouth, a lead-sheathed vessel carrying ingots stamped with the emperor Domitian's name. Other wrecks left blocks of raw marble and a heap of third-century amphora sherds. The moles had become shoals, and ships were hitting them on the way in.

There was at least one attempt to fight back: at the start of the sixth century an embankment of rough undressed stone was tipped into the harbour mouth to salvage the sunken port. It did not take, and by that century the harbour was unusable. Roman Ports cites a 2006 paper in Geology whose title assigns the destruction of Herod's harbour to the tsunami of 13 December 115, while the text of both that article and Wikipedia's says only that seabed deposits record a tsunami in the first or second century. Afterwards the sea front was scavenged, and Caesarea's Muslim rulers and then the Crusaders robbed columns from the Roman ruins to raise a sea wall, a square tower and a jetty in the middle basin.

There is a modern coda, and it is the best argument the site makes for Herod's engineers. The present anchorage was built after 1948 for the fishermen of neighbouring Kibbutz Sdot Yam, using its own artificial island for foundations and the concrete-filled hull of an old immigrant ship sunk at the north end of the new breakwater. The diving club records that this one was falling apart fifty years after it went in.

The most substantial Herodian fabric a visitor can stand inside is behind the quay rather than on it. Fourteen storage vaults, each 7.40 metres high and 21 metres deep, carried the platform of the temple Herod dedicated to Augustus: two wings of four large vaults with six smaller ones between them supporting the monumental staircase up from the pier. Four were restored and fitted out as the harbour visitor centre, which opened in June 2019 after about five years of work, and the parks authority puts a visit at roughly 25 minutes.

The restoration turned up its own story. Only one vault had survived intact, and it was close to collapse. The work then showed that the system had already collapsed once in antiquity and been rebuilt in the Byzantine period, by restorers who kept the Herodian arrangement rather than replacing it. The 2015 to 2019 programme that paid for this cost the Edmond de Rothschild Foundation more than one hundred and fifty million shekels.

Tickets have a quirk specific to this site. The full national park ticket covers everything, but the corporation also sells a cheaper one that admits to the port alone and not to the visitor centre, the theatre or the hippodrome, and a cheaper one again for people coming to eat at the restaurants.

The harbour also keeps producing finds. In February 2015 divers from the local club and marine archaeologists from the Israel Antiquities Authority reported about 2,000 gold coins on the seabed here, a Fatimid hoard more than a thousand years old. Anyone who wants to see the harbour itself rather than its edges needs a mask: the underwater park laid over the drowned structures runs four marked routes and 28 numbered points, taking in the breakwaters, the basins and six anchors from a Bronze Age pierced stone to a modern iron one.