Mamshit National Park

Mamshit is not on the main highland road that ran from Petra to Gaza. The parks authority's brochure places it at a junction instead: the roads linking the Arava (by way of the Scorpions Ascent and Tamara) and the Dead Sea (by way of the Tamar fort and the Tamar and Peres ascents) with the coastal plain and the Hebron mountains met near here. That is a different kind of place from a caravan stage, and it shows in what was built.

The founding date is genuinely unsettled. The parks authority's brochure says a permanent settlement was first established in the middle Nabataean period, in the mid-first century CE, apparently at the top of the hill near where the police building now stands. The Encyclopaedia Judaica entry says about 50 CE. English Wikipedia dates the foundation a century earlier, to the first century BCE, as a trade post between Petra and Gaza.

What is not in doubt is when the town was built up. Most of its buildings, and they are mainly grand private houses rather than public architecture, went up in the second century CE, after Rome annexed the Nabataean kingdom in 106 CE. So the mansions a visitor walks through were raised by Nabataean builders under Roman rule, after the kingdom that made them rich had stopped existing.

The name travelled better than the town. Greek sources call it Mampsis, Arabic Kurnub. Eusebius put it between Hebron and Elath, one day's march from Thamara, and it appears in the episcopal lists of Palaestina Tertia, which is to say it had a bishop.

• At the junction of the Arava, Dead Sea, Petra and coastal plain roads, not on the main Petra to Gaza route
• Permanent settlement from the mid-1st century CE on the parks authority's account
• Most surviving buildings are 2nd century CE, after the Roman annexation of 106 CE
• Known as Mampsis in Greek and Kurnub in Arabic
• Shown on the 6th-century Madaba map as Maps

Mamshit is described almost everywhere, BiblePlaces among them, as the only walled city of the Nabataeans, and that is the fact most visitors arrive holding. The parks authority's own dating complicates it. Its brochure puts the gate and the roughly 900-metre wall in the late Roman period, at the end of the third century CE, and says the wall was widened in the Byzantine period after the earthquake of 363. The circuit is real and it is the only one of its kind in the Negev towns; it is also two centuries younger than the houses inside it.

The gate is worth standing in. It was a passage between two towers with its ceiling carried on three arches, and it is how the town identified itself to the outside world: on the Madaba map, the sixth-century mosaic floor found in a church in Jordan, Mamshit is drawn as a gate with its towers. The parks authority records that the gate was burned and destroyed in the seventh century CE.

The wall also tells you what it left out. Two large complexes outside it, in use from the first to the fourth centuries CE, are read by the parks authority as inns for the commercial caravans: whatever the wall defended, the trade itself slept beyond it.

• Wall about 900 m long, built at the end of the 3rd century CE
• Widened in the Byzantine period after the earthquake of 363 CE
• Gate of two towers and a three-arched passage, burned in the 7th century CE
• The gate and its towers are the city's symbol on the Madaba map
• Two caravan inns stand outside the wall, in use from the 1st to the 4th centuries CE

The largest building here is nicknamed the Nabatu house, and the parks authority gives it about 2,000 square metres of interior courtyards, staircases and rooms. Its capitals still carry small reliefs of a human head, a jug and a bull. In one room the excavators found wall frescoes showing Eros and Psyche, figures out of Greek mythology, painted in a desert town on the edge of the empire. In another they found the coin hoard.

The second great house, called the wealthy house, covers about 1,000 square metres on two storeys, with rooms round a rectangular courtyard: a guard room, a reception hall, a chancery, servants' rooms and a residential wing. A round installation topped with a half dome held the household water jar, the northern rooms have Nabataean capitals at their doorways and wall niches that worked as cupboards, and the stairwell is closed.

The most quietly surprising room at Mamshit is a stable. A typical Nabataean dwelling built against the eastern wall of the Western Church has a courtyard with a cistern and a passage through to a stable that could hold sixteen horses. The parks authority states plainly that raising horses for riding was an important part of the city's economy. English Wikipedia goes further and ties the town's later prosperity to breeding the Arabian horse once the incense trade fell away; the parks authority does not make that claim.

One habit explains the shape of nearly everything here. Nabataean building in the Negev did not use timber, because there was none, so ceilings were carried on stone slabs resting on arches. The tower, originally three storeys high, preserves a ground-floor room with exactly that arrangement, and its stairwell leads to a second storey looking over the ruins and the wadi.

• The Nabatu house: about 2,000 m², with capitals carved as a human head, a jug and a bull
• Fresco fragments of Eros and Psyche on one of its walls
• The wealthy house: about 1,000 m² over two storeys, with a guard room, reception hall and chancery
• A stable for 16 horses in the dwelling beside the Western Church
• Stone ceiling slabs on arches throughout, because there was no timber

The Coin Hoards of the Roman Empire database at the Ashmolean records the find date as 18 August 1966 and the finder as Avraham Negev of the Hebrew University. Its total is 10,320 coins: silver denarii and tetradrachms running from Vespasian to Elagabalus, recovered from a bronze jar. Negev's own account is quoted there: 'While making my regular tour of the site, a worker came up running and presented me with a handful of coins shouting that thousands of them had been discovered in a house which we had started to excavate some time earlier.'

The find spot is as good as the find. The jar had been placed in a strong-box hidden beneath the staircase tower of a Nabataean house, under the fourth tier of steps, facing the inner court of the residential wing. The weight of stone above it had deformed the lid and split the jar along its soldered seam, so more than half the coins had already spread across two or three square metres of debris by the time anyone reached them.

The count is not settled, and a visitor may be told different numbers on the same afternoon. The parks authority says approximately 10,800 silver coins, dating them 75 to 200 CE in the brochure and to the first to third centuries CE on its web page. The Encyclopaedia Judaica says 10,400 Roman tetradrachms; English Wikipedia says 10,500; the Ashmolean database gives 10,320 and notes that 10,343 coins are held by the Israel Antiquities Authority. Its closing reign is Elagabalus, which puts the burial between 218 and 222 CE, roughly a century after the kingdom was annexed. Whoever hid the money never came back for it.

• Found on 18 August 1966 by Avraham Negev, in a bronze jar
• Hidden in a strong-box beneath the fourth tier of a house staircase
• Coins run from Vespasian to Elagabalus, closing between 218 and 222 CE
• Published totals: 10,320 (Ashmolean), 10,400 (Encyclopaedia Judaica), 10,500 (Wikipedia), about 10,800 (parks authority)

The Western Church is named for the man who paid for it. Entered through a colonnaded atrium with a covered cistern under it, it is a basilica of a nave and two aisles, with an apse at the end and rooms for ritual use on either side. The nave floor is mosaic, with geometric patterns, birds, two peacocks and dedications in Greek. The central one reads, in the parks authority's translation: 'God, save your servant Nilus, lover of Jesus, who founded this building. Preserve him and his household.' That inscription is why it is called the Church of St Nilus.

The Eastern Church, part of a monastery, is the more unusual of the two. A flight of steps leads up to an atrium built over a large cistern, and three doors open from it into the church. The mosaic in the nave carries two crosses, which dates it before 427 CE, after which crosses were no longer permitted on church floors. Human bones were found inside, and the rooms flanking the apse held the remains of receptacles for the bones of martyrs, which is where the name Church of the Saints and Martyrs comes from. A large cruciform basin for baptising adults stands beside a small square font for infants.

Both churches went up, on the parks authority's reading, at the beginning of the fifth century CE, and were in use until either the Persian invasion of 614 or the Arab conquest of 636. After that, its brochure says, Mamshit ceased to exist.

The earthquake is legible on the walk. The staircase of the house beside the Western Church still shows the damage of the shock that helped destroy the town, apparently in 363 CE. Korzhenkov and Mazor's archaeoseismic survey, published in 2003, read some 250 damage cases across twelve deformation patterns as the record of two separate events, one around 363 and one in the seventh century, with the shocks arriving from the north. Tali Erickson-Gini later documented a kitchen whose contents were sealed on the floor by the collapse.

How the town ended is still argued. Negev held that Mampsis ceased to matter after the middle of the fifth century; Jodi Magness has documented occupation through the sixth and seventh centuries, with Arabic graffiti on the churches attesting an Early Islamic presence. The signs on site follow Negev.

• Western Church, named from a Greek dedication by one Nilus, with birds and two peacocks in its mosaic
• Eastern Church: two crosses in the nave mosaic date it before 427 CE
• A cruciform basin for adults and a small square font for infants
• Both churches built at the beginning of the 5th century CE
• Earthquake damage of about 363 CE visible on a house staircase
• Korzhenkov and Mazor, 2003: 250 damage cases, two events, shocks from the north

The parks authority is straightforward about this: the first item on its own list of points of interest is 'the remains of the Nabatean city, partly restored'. The clearest case is the market, described on its page as a reconstructed Nabataean street with rows of rooms down both sides that served as shops. Some scholars, the same page notes, read that street as an army camp instead. At Sukkot and Passover the parks authority fills it with an actual market, which is the one moment when the reconstruction does what it was rebuilt to suggest.

Other elements are labelled too. The residential house attached to the Western Church, the one with the stable and the earthquake-damaged staircase, is partly restored. The lower of the Nabataean dams in the wadi was renovated under the Mandate for the camel-mounted police; a second dam downstream is a Mandate structure outright.

The building with the best view is the newest thing on the hill. In 1936 the Mandate authorities put a police station on the highest point of the city, on top of an ancient Nabataean structure, for the camel-back desert patrols that monitored Bedouin movement and, in the parks authority's own wording, prevented Jews from travelling the area with an eye to settling it. Its roof is now the lookout over the ruins, and the building is used for events.

Even the excavation history comes in incompatible versions. The parks authority brochure has S. Appelbaum making trial soundings for the Hebrew University in 1956 and A. Negev excavating in 1956 and 1957, renewing the work in 1989, with T. Erickson-Gini continuing for the Israel Antiquities Authority in 1993 and 1994. The Encyclopaedia Judaica dates Negev's excavations from 1966; BiblePlaces gives 1965 to 1973; the Ashmolean's hoard record gives 1965 to 1967. The site was declared a national park in 1966 and inscribed by UNESCO in 2005 at the committee meeting in Durban.

• The market street is a reconstruction; some scholars read it as an army camp
• The residential house by the Western Church is partly restored
• The lower dam in the wadi was renovated under the British Mandate
• The police station on the summit dates from 1936 and is now the viewpoint
• Declared a national park in 1966; inscribed by UNESCO in 2005