Herod's Northern Palace

Herod's engineers had a flat summit roughly 650 metres long to build on, and they put the king's own quarters on the one part of it that runs out of ground. The plateau narrows to a point at its northern end and then falls away; the palace begins on that point and continues down the face, on ledges squared off and walled up to make three usable platforms. The parks authority calls it a daring structure built at the edge of a deep ravine, with strong retaining walls doing work that is invisible from the top and obvious from below.

There is a plain reason for choosing the north end. It is the shaded side, out of the worst of the sun, and it faces the escarpment and the Dead Sea rather than the desert. Josephus says the palace enjoyed better conditions and a view running as far as Ein Gedi: the site was picked for climate and outlook, not for defence.

The palace was also cut off from the rest of the mountain. A high white plastered wall ran across the neck of the summit south of it, leaving only a narrow passage at the eastern end for the staircase down: anyone on the plateau, garrison included, was on the far side. Josephus places Herod's palace "at the western ascent" while adding that it was "within, and beneath the walls of the citadel: but inclined to its north side", a description that has sent readers looking in two directions ever since.

It went up in stages. Ehud Netzer identified three construction phases at Masada; on Stiebel's summary a building on the upper terrace belongs to the first, and the palace as a complex, with the large bathhouse and public storerooms south of it, to the second.

The upper terrace is the residence. Four rooms and a central hall sit south of a semicircular balcony that was once ringed with columns and is carried on a double wall built out from the terrace edge. The rooms were paved with geometric mosaic and the walls and ceilings painted. In the southwestern room the parks authority records a black and white mosaic of hexagons, a pattern it calls typical of Rome and its surroundings and reads as a clue to where the craftsmen took their ideas from. The balcony looks out over the Ha-'Etekim escarpment, the Dead Sea and the Roman siege system.

The middle terrace, about 20 metres down, is the hardest to read on the ground because almost nothing of the superstructure is left. What survives is two concentric circular walls, which served as the platform for a round columned building; the parks authority is blunt that only the foundations remain. Stiebel's entry describes a staircase on the west, a large room with traces of frescoes on the east, and a roofed colonnade between them which he suggests may have been a library. It is the one part of the palace whose architecture he calls completely Hellenistic rather than Roman.

The lower terrace, roughly 15 metres below that again, is the best preserved of the three. Great supporting walls raise a nearly square platform on a spur that tapers to a point; low walls around it formed porticoes on all sides. The outer walls carried stucco, and the inner walls frescoes painted to imitate slabs of coloured marble and geometric panels. In the eastern corner is a small bathhouse in the Roman manner, and it was here that Yigael Yadin's expedition found three human skeletons.

Stepped ritual immersion pools were cut into the lower terrace and beside the middle one during the revolt, not in Herod's time: the rebels were living in the king's private rooms a lifetime after he died.

• Upper terrace: four rooms and a central hall, mosaics, painted walls, a semicircular colonnaded balcony on a double wall
• Middle terrace: two concentric circular walls carrying a round columned building, foundations only
• Lower terrace: a colonnaded hall on a raised platform, frescoes imitating marble, a small Roman bathhouse
• Ritual immersion pools added in both lower levels during the revolt

The palace is not stylistically consistent, and that is the interesting thing about it. Stiebel's Encyclopaedia Judaica entry treats it as a case study in the moment when Roman taste displaced Hellenistic taste in the eastern Mediterranean: the upper terrace was built in Roman style, the middle terrace is Hellenistic throughout, and the lower terrace was fitted with a Roman-style bathhouse. One building, two architectural worlds.

The Roman connection may be closer than a matter of fashion. Gideon Foerster argued that the plan of the upper terrace was directly influenced by the villa in the grounds of the Farnesina in Rome, a house attributed to Marcus Agrippa, who was Herod's patron and personal friend. If that holds, the drawing on the desk at Masada came from the circle of architects who built for Augustus's closest colleague.

The columns are the detail most visitors walk straight past. On the lower terrace they look like tall monolithic shafts of fluted stone. They are not. They are drums of soft sandstone stacked up, then plastered and fluted so the join disappears and the whole thing reads as one piece. The trick is ancient, not modern: this is what Herod's masons did in a place with no marble and no way of hauling it up a cliff. Josephus, who says the buildings were supported by "pillars of single stones", appears to have been taken in by it.

The same stone carries a signature of sorts. Stiebel notes that mason's marks in Hebrew letters are visible on all the pillar drums at Masada, indicating where the stone cutters came from, while the softer stone used for drums and capitals was brought in rather than quarried on the summit.

What the palace was for is still disputed. Yadin read the Western Palace as Herod's ceremonial building; Netzer rejected that and gave the ceremonial role to the Northern Palace. Nothing found so far settles it, and the two readings put the king in different rooms.

The parks authority's own description of the palace lists "mosaic floors and restored wall paintings", and the word restored is carrying weight. The paintings on the lower terrace survived because the building fell in on them: buried under its own collapse, plaster keeps its colour in a way that exposed plaster in the Judaean Desert does not. Conservation here has been continuous since the excavations, and the parks authority describes decades of investment in conserving the site alongside the interpretive panels and the access works.

How far that conservation goes is easiest to see next door. Frescoes lifted from the commander's residence, the building at the southern entrance to the Northern Palace complex, spent roughly two decades in storage after Yadin's dig because they were deteriorating outdoors. They were restored by the Italian conservator Maurizio Tagliapietra and put back in 2014, in positions worked out from photographs taken during the excavation. Nothing on Masada that looks finished should be assumed to have stayed that way since 1965.

On the terraces themselves the split between ancient and modern is uneven. The lower terrace keeps its frescoed walls and the footprint of its porticoes. The middle terrace has lost effectively everything above its two foundation rings, which is why it reads as a pair of circles in the ground rather than as a building. On the upper terrace the balcony survives as the double wall that carried it; the columns that stood around it are gone. The steps you walk down are not the Herodian staircase, though parts of the original survive on the middle and lower levels.

• Original: the frescoed lower-terrace walls, the mosaic in the southwestern room upstairs, the retaining and foundation walls
• Gone: the columns of the upper terrace balcony and almost everything above the middle terrace foundations
• Restored: wall paintings, described as such by the parks authority
• Modern: the visitor staircase and its railings, and the interpretive models on the plateau

In late November 1963, in the small bathhouse on the lower terrace, Yadin's expedition uncovered the remains of three people, along with a woman's scalp with the braids still in it and a pair of leather sandals. Yadin announced them at a press conference as the bones of Masada's Jewish defenders, and no other single find did as much to fix Josephus's account in the public mind. The braids are displayed at the Yigael Yadin museum at the foot of the mountain.

The identification has never been secure. Stiebel's encyclopaedia entry sets out the problem plainly: Yadin claimed the three as a rebel family, but the condition of the skeletons and the absence of many protein-rich bones may instead indicate that the bodies were dragged there by hyenas. That is a different story about how three people ended up in a bathhouse, and it is not a story about a last stand.

A separate group of 25 skeletons had been found a month earlier in a cave on the southern cliff, and those are more contested still. Pig bones were found among them, which led the anthropologist Joe Zias to argue they were Roman rather than Jewish dead; Zias has published a recollection of Yadin telling him, of the cave remains, "I never said that all the remains found there were Jewish."

The remains were nonetheless given a state funeral. On 7 July 1969, 27 skeletons, three from the Northern Palace and 24 from the southern cave, were buried near the foot of the Roman ramp with full military honours, the army's chief chaplain Shlomo Goren officiating and Yadin reading from Josephus. What was excavated as an open question was reburied as a settled one.

The palace is not on the way to anywhere. The top terrace is at the northern end of the plateau, and everything below it is a descent and a re-ascent by the same steps. The parks authority describes that route as many steps with railings on both sides, and offers the models and lookouts on the plateau as the alternative for anyone who would rather not. That is an official acknowledgement that the two lower terraces are optional.

Nothing about the palace is separately ticketed or separately timed. It opens and closes with Masada National Park, which the parks authority lists as 08:00 to 17:00 in summer and an hour earlier through the winter, closing an hour earlier again on Fridays and holiday eves. The last cable car goes up an hour before closing, so the last sensible moment to start down is well before the gate shuts.

The order of the day matters more here than at most of the plateau complexes. The steps are exposed, the return leg is uphill, and Masada's paths run on a heat schedule rather than a timetable: in extreme summer conditions the parks authority puts the last ascent on the Snake Path at 07:00. Doing the palace first, straight off the cable car or the top of the path, buys shade on the north face and an empty staircase.

The view is the other argument for going down. From the balcony and the terraces below it the escarpment, the Dead Sea and the outlines of the Roman camps on the marl plain line up in one frame, which is true from nowhere on the summit.