Taninim Stream Nature Reserve

By the late Roman period Caesarea had outgrown the high-level aqueduct, which had taken every spring within reach that stood at a usable height. The nearest large untapped source was the cluster of springs in the Kebara basin, near what is now Kibbutz Ma'agan Michael. The catch was elevation. Those springs rise between three and a half and four metres above sea level, and the water had to reach roughly six metres to run into Caesarea at all.

The answer was to stop treating the marsh as a marsh and turn the whole of it into a header pool. Two dams of large ashlar went in. One, the wall standing in this reserve, blocked the breach that Nahal Taninim and its tributary Nahal 'Ada had cut westward through the kurkar ridge to the sea. The second, about 2 km north and far longer, closed the basin at its northern end. Between them, the Carmel and the ridge, the springs filled a sheet of standing water the parks authority puts at about 6,000 dunams.

What came out of it is easy to underrate. Published calculations give the low-level aqueduct, the one starting at this dam, a capacity of some 2,560 cubic metres an hour: more than the high aqueduct's first channel and its Hadrianic second channel together. The two supplies differed in kind as well. The high aqueduct carried spring water of drinking quality from the upper valley at Shuni; this one carried lake water, brackish and full of whatever the basin held.

The figures are not quoted identically everywhere. The parks authority gives the dam as 194 metres long and the lake as about 6,000 dunams; the Hebrew accounts of the Caesarea aqueducts and the iNature reserve entry round the dam to 200 metres and put the reservoir nearer 5,000 dunams, six metres deep.

• Springs rise at about 3.5 to 4 m above sea level; the water had to reach about 6 m to flow to Caesarea
• Two dams: this one across the ridge, and a second of about 900 m to the north
• Low-level aqueduct capacity: about 2,560 m³ an hour
• High-level aqueduct, for comparison: about 930 m³ an hour in Channel A and 1,630 in Channel B

The dam was excavated by the Israel Antiquities Authority between June 2001 and August 2002, under 'Abed a-Salam Sa'id and Uzi 'Ad, with the parks authority and funding from the Ministry of Environmental Affairs and the Carmel Drainage Authority. The dig was a preliminary to conservation and public opening. The excavators dated the wall to the end of the Roman period and the beginning of the Byzantine; the parks authority puts it at the start of the fourth century CE at the latest.

Very little of what you stand on is ancient. Across the crest, everything down to between 0.7 and 1.5 metres below the present walkway proved to be Ottoman fill, with almost nothing of the original top beneath it. The core is rubble and reused ashlar bonded with grey mortar, most of it washed out where water got in. The eastern face was cast against wooden formwork that left its impressions, and patches of pink Byzantine hydraulic plaster survive there. One trench went to over two metres below sea level without reaching the bottom of the foundation.

The wall is not a uniform thickness either, and the parks authority explains why: supports were added over the centuries to stop it collapsing. One find dates it from underneath. A rock-cut burial cave at the northern end of the foundation was sealed when the dam went up, and held three limestone sarcophagi of a Samaritan type common in the second and third centuries CE.

The most conspicuous features are reconstructions and are worth recognising as such. The wooden gates in the regulating installation were rebuilt, and raising them releases water from the pool behind. The parks authority calls the full reconstruction of an ancient dam to recreate its lake one of the wonders of the site, and says it may be unique in the world. That is the operator's assessment of its own work, but it describes an unusual decision: most excavated waterworks here are conserved dry.

Below the western foot of the dam the kurkar is cut away into a hillside of quarried chambers, and those chambers are flour mills. Six were hewn in the Byzantine period; seven more were built in the late Ottoman period, running 13 milling stations between them, two of them with four stations each.

The Byzantine six are the reason archaeologists come. Their feeder channels carry circular scour marks, and from those marks the excavators reconstructed the wheel that made them: between 1.9 and 2.3 metres across, and vertical, mounted so the water strikes its blades on the way past. The parks authority states that a Byzantine vertical wheel of this kind is not known anywhere else in Israel, and that the type is otherwise recorded here only from the Crusader period. The Ottoman mills use the commoner arrangement: a built and plastered vertical shaft about 1.3 metres across, dropping water onto a horizontal wheel at the bottom.

The two systems were never in use together. The Byzantine feeder channels were deliberately blocked in the Ottoman period, apparently to clear access to a newer mill, and the fill sealing them dates the closure.

Several wheels have been reconstructed, and the small lake retained beside the dam holds enough water to turn two of them. The parks authority is explicit that only reserve staff operate them, so a turning wheel is something a visitor may see and cannot arrange.

The strangest thing in the reserve is a channel that never carried water. Running south from the dam, the excavators traced a rock-cut aqueduct for 390 metres: a narrow slot 45 to 50 cm wide at the bottom and up to 2.5 metres deep, with ledges along the top for covering slabs. Where it met the low-level aqueduct to Caesarea, that later channel rode piggyback on it for 145 metres, and at the point of divergence a wall was built inside the older cutting to push its water across into the Caesarea line.

Then the excavation followed the abandoned channel past the junction and found the quarrying had never been completed. Sa'id and 'Ad concluded that this aqueduct never functioned. The parks authority offers the reason: its builders worked out that the fall between the head of the channel and the city was not enough to move water to Caesarea, and stopped.

The low-level aqueduct that replaced it carries a puzzle nobody has resolved. Excavated for 320 metres from its outlet in the distribution pool, its floor sits about half a metre lower at the northern end than at the southern, which is the wrong way round for water heading south to Caesarea. The excavators wrote that the reason for this reverse gradient is still unclear. The standard account of the Caesarea aqueducts says something equally odd: the low channel has no gradient at all, and moved its water on the strength of its cross-section.

The unfinished cutting has ended up as the reserve's best half hour: a wet route 260 metres long in about 45 cm of water, and the only place the parks authority permits anyone into the water at all.

The reservoir long outlived the empire that built it. It held for roughly a thousand years before silt and neglect turned the standing water into the Kebara marshes, a breeding ground for anopheles mosquitoes and a focus of malaria on the coastal plain. The people who lived off it were the Arab al-Ghawarina, who fished the marsh, grazed water buffalo on it and wove mats from its papyrus and reeds.

Draining it was a Rothschild project. The Jewish Colonisation Association bought the marshland in 1922 with the baron's money, put around a quarter of a million pounds into the works and brought in a French engineer, Léon Cron, who had drained swamps in Algeria. PICA, which took the association's place in 1924, bought replacement land for the Bedouin and in 1926 built them Jisr az-Zarqa on the kurkar ridge, the town whose roundabout you now turn at to reach the reserve car park.

The engineering was rescued by a labourer. Dov Kovlanov arrived as an ordinary worker, turned out to be a Russian agricultural engineer specialising in drainage, judged that Cron's plans would not solve the problem and submitted his own. They went to Paris, were approved, and he took over. His method was to find the thousands of underground seepages feeding the marsh and pipe them: brackish water to the sea, fresh water to the fields. That took some 50 km of clay pipe, imported from France because local clay would not make it.

The crocodiles went out with the marsh. Henry Baker Tristram recorded the capture of one over three and a half metres long in the Zarqa in 1864 and thought it the only crocodile ever taken in Asia; a female of about three metres with 48 eggs in her was caught in 1877. The last was hunted in 1912 and still exists, preserved in Father Ernst Schmitz's zoological collection, now held by the Steinhardt Museum of Natural History at Tel Aviv University.

The reserve is small and long, running along both banks from the coastal highway to the sea. iNature gives its declared area as 348.55 dunams, accumulated in stages: two blocks totalling 55.3 dunams in the summer of 1972, an extension of 211 dunams in December 1992 that carried it down to the shore, and a redesignation in December 2015 turning about 11.5 dunams into national park so the entrance buildings could stand there legally.

The approach matters, because the obvious one is shut. The old access from Kibbutz Ma'agan Michael to the north is now closed by a gate; the working entrance is reached from the old Haifa to Tel Aviv road, over the bridge above the coastal highway towards Beit Hananya, then round to the roundabout at Jisr az-Zarqa. Without a car, bus 68 from Hadera stops about 350 metres south of the ticket office.

Two things worth walking to sit outside the main circuit. The stream mouth carries the remains of a stone bridge built at the end of the Ottoman period for Kaiser Wilhelm II's visit in 1898, and immediately south of it stands Tel Tanninim, the Krokodeilon polis of Strabo and Pliny, dug between 1996 and 1999 by Robert R. Stieglitz of Rutgers University. Its Byzantine and Crusader names both mean saltworks tower.

The stream repays a slow walk. Yellow water lily and hornwort grow in the channel, both uncommon, and the salt-tolerant plants on the banks are there because the water is brackish. Soft-shelled and marsh turtles, catfish and tilapia live in it, herons nest in the eucalyptus stand between Taninim and 'Ada, and the hillside north of the old dam flowers heavily with autumn narcissus in mid-November.