I first learned of Sepphoris’s ancient water system in 1975 from a local resident named Buki. He told me about a huge underground cavern three stories high and stretching more than two football fields long. It sounded to me a little like a flight of fancy. “When I see it, I’ll believe it,” I said. So Buki led me down a steep slope into the middle of an enormous hall 30 feet high, 625 feet long, 10 feet wide and capable of holding more than a million gallons (160,000 cubic feet = 4,300 cubic meters)—enough to supply water to 15,000 people for 2 weeks.1
When Buki told me that no one had ever fully investigated the reservoir, I took his words as a challenge. At the time, I was on the staff of the Israel Defense Forces’ Department of Israeli Studies, which was devoted to teaching soldiers about the land and history of our country. For the next 10 years, I surveyed the region east of Sepphoris. By 1985, I had mapped the 8.5-mile system of tunnels, channels and aqueducts that brought fresh water to the city from the first to seventh century C.E.
Bringing water to a city on a hill like Sepphoris, which rises 250 feet above the surrounding valley, is no easy task. That is why the earliest residents of the town simply relied on rainwater collected in cisterns. But by the first century C.E., this water supply proved insufficient for the growing city.
Not far from the foot of the hill of ancient Sepphoris lie copious springs (the Sepphoris Springs) that today irrigate the surrounding valley. But despite their proximity to the town, the springs were of marginal use to the ancient residents, who would have had to use pack animals to haul the water up the side of the hill. To channel water into the city, they needed to find a spring elevated higher than Sepphoris, so that the force of gravity would conduct the water across the surrounding valley and up the slope into the city.
The closest elevated springs lie further east, near the foot of Mt. Yona, just north of Nazareth—about 4 miles southeast of Sepphoris. Which meant that the city dwellers had to build a fairly extensive aqueduct if they wanted to take advantage of this fresh water supply. And so they did.
In the first and second century C.E., the residents of Sepphoris built not one but two aqueducts, which began at two different points near the foot of Mt. Yona, came together briefly and then diverged. The northern aqueduct emptied 036into a pool just outside Sepphoris; the other led to the cavernous reservoir Buki had shown me and then continued into the city proper.
Examining the literature, I learned that I was not the first to investigate the Sepphoris water system. The reservoir was initially published in 1866 by the British officers Charles Wilson and Charles Warren, and was further researched in 1872 by the London-based Palestine Exploration Fund (PEF), which was then mapping the Holy Land. The PEF team also discovered the sources of water for the northern aqueduct (the Amitai and Genona springs). They traced this system for about a mile, identifying a collapsed tunnel and a wall that supported the water channel as it crossed a small valley.
In 1931 the University of Michigan team that first excavated ancient Sepphoris also surveyed the aqueducts, but only for a half mile east of the city. Their efforts were halfhearted at best. Of the water source, they wrote simply: “The aqueduct begins somewhere to the east.” An account in their 1937 excavation report suggests that they were looking for almost any excuse to curtail their investigation: “A number of cisterns on the Tell were cleaned with the hope of finding some connection between them and the aqueduct. There was a promising one with steps cut leading down to it. After considerable trouble, work on it had to be abandoned when a serpent fell into it and the men were too frightened to continue the excavation.”2
Finally, in 1966, the Israel Department of Antiquities (now the Israel Antiquities Authority) uncovered two further sections of the aqueducts (although they mistakenly concluded that one of the sections did not lead to Sepphoris).
Until our project began, however, no one had made a systematic study of the entire aqueduct system from its sources in the nearby mountains to the city.
Aqueducts were built as early as the ninth century B.C.E. (in the Assyrian Empire, in modern Iraq), but it was in the Roman period that they became widespread, to such an extent that the structures became almost a symbol of the Roman Empire. In the first century C.E., the water commissioner of Rome, a man named Frontius, praised his city’s nine aqueducts, which covered 264 miles: “With such an array of indispensable structures carrying so many waters, compare if you will, the idle Pyramids or the useless, though famous, works of the Greeks!”3
The renowned first-century B.C.E. Roman engineer and architect Vitruvius, in the oldest extant Western treatise on architecture (De Architectura), offers 037explicit instructions for building an aqueduct, including finding a spring, choosing an aqueduct design suitable to the local topography, measuring the angle of descent, bridging valleys, tunneling through mountains, selecting the best spot for a reservoir and distributing water within the city.a As the Roman Empire expanded, Vitruvius’s treatise encouraged many an engineer throughout the Near Eastern world to convert theory into practice. By the first century C.E., nearly every Roman city had some sort of aqueduct.
Today, the term aqueduct conjures up images of multitiered arched bridges that carried water at the height necessary to reach Roman cities built on hills. But the term, which simply means “water duct,” also refers to much humbler man-made channels cut into bedrock, shallow aboveground channels constructed of stone and roofed with stone slabs, and underground tunnels. It is these more modest, yet extremely efficient, types of aqueduct that we find at Sepphoris.
The oldest Sepphoris aqueduct is dated by pottery and plaster to the early first century C.E., perhaps to the reign of Herod the Great’s son Herod Antipas (4 B.C.E.–39 C.E.). This northern aqueduct was fed by the Amitai and Genona springs, near the modern village of Mash-had. These springs do not supply a great deal of water, but they are located high enough in the hills (1,570 feet above sea level) to provide the necessary slope down to the city (950 feet above sea level). The first mile of the Mash-had aqueduct descends rapidly from the hills. Here the aqueduct is cut directly into the bedrock, which is why we can still see some poorly preserved sections of it today, starting about 300 yards from the spring (photo, p. 36). Gray plaster, made of ash mixed with lime, lined the walls of the 12.5-inch-wide water channel.
As the population of Sepphoris grew in the second century C.E., a second aqueduct was built, starting at the el-Qanah spring, today in the village of e-Reina. The abundant e-Reina spring supplied nine times more water than the Amitai and Genona springs at Mash-had. We do not know why the first-century C.E. builders did not use the el-Qanah spring in the first place. Perhaps they simply did not recognize that this spring, which is 470 feet lower in elevation than those at Mash-had, was of sufficient height to serve as a water source for Sepphoris.
A second-century C.E. springhouse built of well-cut ashlars (rectangular building stones) still protects the el-Qanah spring. But the aqueduct itself has been badly damaged by heavy tractors and land clearing for modern farms. To reach Sepphoris, the e-Reina aqueduct must have crossed the stream-bed of Nahal Zippori (Valley of Sepphoris) on an arched bridge before becoming a tunnel about a half mile north of the valley. The local Arabic name of the valley, Wadi el-Gisser, meaning “Valley of the Bridge,” reinforces this theory. Unfortunately, nothing remains of the bridge.
About 3 miles from the springs, the e-Reina and Mash-had aqueducts joined to form a single raised channel, supported by a foundation wall, for about a half mile. One mile outside the city, the aqueducts diverge into two aboveground channels. One channel, the stone roof slabs still partially preserved, continued for another half mile, to the outskirts of Sepphoris, where it debouched into an open pool some 70 feet long by 50 feet wide and at least 11 feet deep. The size of the pool, its proximity to the city, its smoothly plastered surface, and its 16-foot thick walls, which provided a wide area around the pool for recreation, indicate it may have served as a swimming pool, where the Sepphoris residents refreshed themselves during the warm spring and early summer months, when abundant water flowed from the springs.
The second channel, still well preserved in places, ran only for about 300 feet, to the cavernous reservoir, the most impressive feature of the entire water system. 038Local legend tells of a runaway cow that entered the reservoir near Sepphoris and resurfaced 10 miles away, in Sheikh Abreq (today Beth-She’arim). The Sepphoris system was even mentioned in the Mishnah, the earliest code of Jewish law, which was prepared in Sepphoris around 200 C.E. In a technical discussion about drawing water on the Sabbath, Rabbi Judah observes: “It actually happened with the aqueduct, which flowed from Abel to Sepphoris, that water was drawn from it on the Sabbath on the authority of the Elders.”4
From 1993 to 1994, I led a team from Tel Aviv University’s Institute of Archaeology that extensively excavated and restored the reservoir so that in the summer of 1995 we could open it to visitors. It is now one of the premier attractions of the Sepphoris National Park.
The long, narrow reservoir was hewn on three sides from bedrock—a soft, chalky limestone called kirton. (The north wall is formed of hard limestone—a difference caused by a geological fault that crosses the reservoir.) The reservoir was constructed at the closest point to the city where kirton bedrock of the appropriate depth could be found.
Our goal was to determine how water entered the reservoir and then exited to points within the city. Having detected a small cut in the bare rock aboveground near the eastern end of the reservoir, we began digging. We quickly exposed the end of the water channel. The channel poured into a 16-foot-deep sedimentation basin that trapped silt so that only clean water passed into the reservoir, through a chute 3 feet above the basin floor. A large number of ceramic vessels found on the floor of the sedimentation basin date to the seventh century C.E. Apparently they fell in or perhaps were thrown in when the system went out of use.
The walls of the reservoir are lined with two layers of plaster, an earlier, light pink layer dated to the first or second century C.E. and a second, dark pink layer that was applied over fourth-century C.E. pottery fragments, indicating that it dates later than this.5 It may have been applied when the reservoir was repaired and enlarged after an earthquake devastated the Galilee in 363 C.E.
Nine openings in the ceiling provided access to the reservoir. The steepness of the shafts suggests that they were originally used to haul hewn material straight up from the reservoir as it was being constructed. Two of the openings have staircases that allowed people to clean and maintain the interior.
Once we knew how water got into the reservoir, we were eager to discover where it flowed out. With the help of a small Bobcat tractor, we began clearing the 040system’s westernmost entrance. We unearthed steps leading down to a small opening in the floor of the reservoir. The opening led into a narrow, 180-foot tunnel that had been sealed off for some 1,300 years. Crawling on our hands and knees, we came almost immediately to an area where we could stand upright. Along the side walls were small projections—supports for oil lamps, used by the diggers who built the system. Soon the passage grew smaller, and we were reduced once more to crawling until, finally, we had to stop because we had reached a wall. We had found the end of the reservoir!
Inserted in the wall was a single lead pipe, 4 inches in diameter. This pipe, when opened, provided the sole egress for water passing through the reservoir into the city. We were, of course, eager to discover the other end of this pipe behind the wall. But this would require several months’ work.
The aboveground aqueduct system reappears about 650 feet west of the reservoir and continues into the city’s residential quarter. This last stretch of the aqueduct is cut into a shallow foundation of natural rock. The stone slabs covering the conduit survive in their entirety. But what lay in between the reservoir and this final surface channel?
Suspecting that the pipe emptied into an underground tunnel, we scoured the area on foot. We detected a small hole in the ground, which turned out to be the entrance to a steeply angled shaft with an opening at the bottom. Equipped only with a flashlight, a brave member of our team, Arik Rosenberger, headed in. We waited nervously until he emerged about 15 minutes later and told us of his trip down a twisting 150-foot-long tunnel with a water channel running along the floor.
After long and arduous work clearing and excavating this area, we discovered that the shaft we had found was one (later called shaft 3) of six shafts leading into the tunnel. The shafts were apparently used by the workers who dug the tunnel. Two teams of tunnel diggers would enter a shaft. At the bottom, they would split up and dig in opposite directions towards the adjacent shafts. Meanwhile, two teams of workers who entered the adjacent shafts would be digging toward them. From the variations in the tunnel walls between the shafts, we could identify four places where the teams of 041cutters had met. This tunnel, which was completely unknown before our work began, measured 770 feet in length.
We cleared out the tunnel leading from shaft 3 to shaft 2, and finally to shaft 1. At the bottom of shaft 1, we found the other end of the lead pipe. To measure the ancient pipe, we inserted a flexible plastic pipe. It was 20 feet long.
A metal lug was fastened solidly to the floor near this end of the pipe. The lug was apparently part of a gate valve, or stopcock, which was opened and closed to control the passage of water. One of the water system staircases leads directly down to the pipe so that a worker could easily climb down to turn the stopcock, thereby adjusting the supply of water flowing through the lead pipe. Pressure would force the water through the tunnels and channels that led into the city.
At last, we had mapped the whole system.
We still don’t know precisely where the main line of the aqueduct system ends inside the city. We know it reached only about four-fifths of the city. This does not conform to the standard Roman practice of directing water to the city summit. Perhaps Sepphoris lacked the financial or technological resources to build the aqueduct bridges necessary to bring the water up the hill. Excavators in the lower city have not yet found the precise spot where the aqueduct ended. But the discovery of numerous small water channels and a section of lead pipe near a fifth-century C.E. building famous for its exotic mosaics of the Nile River suggests they are close.
I first learned of Sepphoris’s ancient water system in 1975 from a local resident named Buki. He told me about a huge underground cavern three stories high and stretching more than two football fields long. It sounded to me a little like a flight of fancy. “When I see it, I’ll believe it,” I said. So Buki led me down a steep slope into the middle of an enormous hall 30 feet high, 625 feet long, 10 feet wide and capable of holding more than a million gallons (160,000 cubic feet = 4,300 cubic meters)—enough to supply water to […]