During June 2003, new excavations began at the Roman fort of Yotvata (‘Ein el Ghadyan; License No. G-17/03; map ref. NIG 2043/4217; OIG 1543/9217), some 40 km north of Elat. The excavations were sponsored by the Jewish community of Toronto, the Elot Regional Council, the ‘Arava Institute for Environmental Studies at Qibbuz Qetura, and the Foundation for Biblical Archaeology (North Carolina, USA), and directed by U. Avner of the ‘Arava Institute for Environmental Studies, G. Davies of Florida International University, and J. Magness of the University of North Carolina at Chapel Hill. Assisting staff members were N. Bierling (drafting and photography) and T. Levine (registration and pottery restoration). The volunteers were mostly local qibbuzim members, as well as staff from the Elat Field School and high-school students. This was the first in a series of five projected seasons of excavations at the fort.
The Roman fort was badly damaged in 1958, when a large trench for an oil pipeline was cut through its center from northwest to southeast. Z. Meshel of Tel Aviv University conducted trial excavations at the fort in 1975–1976, consisting of a number of soundings. The most important finding from the site was a monumental stone inscription discovered accidentally in 1985, which presumably was set above the main gateway of the fort. The inscription identified the fort as a Tetrarchic foundation (c. 300 CE).
The plan of the fort is a standard quadriburgium––a roughly square structure with four projecting corner towers (external dimensions 39.4 × 39.7 m). The towers are oriented toward the four cardinal points. However, for the sake of clarity we refer to the south corner tower as the southeast corner tower, the southeast wall of the fort as the east wall, the southwest wall of the fort as the south wall, and so on. All site plans show true orientation relative to the cardinal points. The main gate (with the monumental inscription) was located in the center of the eastern wall. The outer walls had a stone foundation (height c. 1.4 m above ground level), with a mud-brick superstructure. Most of the stones were not hewn, and some were roughly cut ashlars. The gate to the fort and entrances to the towers were built of ashlars, sometimes incorporating finely dressed Nabatean ashlars. The interior partition walls were constructed from mud bricks. It is likely that the fort had a second story.
During this first season attention was focused on the southeastern side of the fort. An area (4.0 × 9.5 m) was opened along the inner side of the fort's south wall, up to and including the entrance to the southeast corner tower. As in Meshel's soundings, two main occupation phases separated by a debris layer (thickness c. 1 m) were discovered. However, our excavations revealed that the upper occupation phase was significantly later than what Meshel had suggested, dating to the seventh century instead of the fourth century CE.
The Upper Occupation Phase (Fig. 1)
The site is covered by a very hard crust (thickness 0.1–0.3 m), which was apparently formed due to the crystallization of salt and gypsum. Immediately below the crust remains of mud-brick walls came to light (Fig. 2). These formed one main room in the center of the area (L2001). A row of small rooms or cells (from east to west Loci 2003, 2005, 2006), enclosed with mud-brick partition walls (thickness 0.20–0.35 m), abutted the inner face of the fort's south wall. Part of another room (L2002) occupied the entire western side of the area and was separated from L2001 by a mud-brick wall (W502; thickness 0.45–0.48 m). After it was realized that a mud-brick wall (W507) of the same thickness as W502 ran along the north side of the area, the trench was extended 1.5 m northward. Another mud-brick wall (W504) that was parallel to and east of W502 cut through the center of L2001, abutting the south wall of the fort, but terminating before it reached W507. It turned out that W504 was built on top of the remains of an earlier staircase (see below).
A large pit (L2009, L2010; diam c. 2 m; depth 0.5 m) that was dug through the upper occupation level in the middle of L2001 and extended into L2005 provided definite evidence for recent presence of Beduins. It contained wool fabric (probably from a tent), fragmentary leather sandals and glass bottles.
A cell (L2006) on the southwestern side of the main room (L2001) contained the remains of a hearth that abutted the south wall of the fort. It was defined by two parallel clay piers that gave it the shape of a horseshoe. The inner faces of the piers were reddened from intense heat, and the area inside them was filled with thick ash layers. A shallow depression immediately in front of the hearth contained blackened soil and a large, discrete chunk of charcoal. There is no evidence that this hearth was used for metallurgical activity.
The floor level of the upper occupation phase was detected at an elevation of c. 77.5 m. Large quantities of restorable ceramic vessels were found on the floors of the rooms, including the complete base (toe) of an amphora in L2002, which had a hard, white plaster or lime floor. Elsewhere, the floors consisted of packed beaten earth.
The Lower Occupation Phase (Fig. 3)
The lower occupation level was divided into two main areas flanking W504 of the upper phase: L2018 on the east and L2019 on the west. The floor of this phase was reached at elevation of c. 76.5 m. The main architectural feature was a staircase that projected from the fort's south wall to the north, into the center of the excavated area (Fig. 4). When the fort was reoccupied during the upper phase, the base of the staircase was partially covered by W504. The bottom steps consisted of four large, smooth and nicely cut ashlars, which had subsided and tilted to the north. The staircase must have made a turn to the south at a landing under the continuation of W504. Below the presumed location of the landing and under the base of W504's east face (in L2018) was a square earthen platform covered with terra cotta tiles that were set in mortar. Several coins were discovered around the bottom of the steps and on the floor around the earthen platform. Under the base of W504, along its west side (in L2019), another, larger platform was built of small stones compacted by mud bricks. It is apparently the foundation or support for the staircase after it turned at the landing and continued to the south. A collapse from a second story or perhaps from the roof was discovered above the floor along the south side of L2019. It contained large fragments of pottery vessels at an elevation of 76.7–77.0 m. The quantity of pottery fragments and the manner of its deposition, as well as the size and sturdiness of the staircase suggest that the fort was originally two stories high.
On the eastern side of L2018 a rectangular installation or pit enclosed with a low, thin mud-brick and stone curb (L2015; 0.55 × 1.32 m) was found. This pit appears to have been used for the dumping of refuse, as it contained numerous animal bones, potsherds and large quantities of charcoal, including one very large chunk.
All the floors, including the entrance to the southeast corner tower, consisted of packed earth with lime or plaster flecks. Most of the finds from the lower level, including coins, were discovered on top of the floors. The fills that accumulated above the floors contained numerous fragments of plain white wall plaster, lying face down as they had fallen from the walls. No signs of conflagration or other violent destruction were observed at the lower occupation phase and its floors did not yield restorable pottery vessels, indicating that this phase ended with abandonment.
To the west of L2002 and outside the excavated area, one of Meshel’s soundings (L1512; our L2013/L2016) was cleaned. The remains of a north–south wall or, more likely, the foundations of a wall (W509) were discerned at the current ground level, on the western side of the sounding. This wall, which was noted by Meshel, was apparently badly damaged when the pipeline was laid. Once cleaned, the eastern side of the sounding provided a good record of the stratigraphic sequence. The white plaster floor of L2002 is visible at an elevation of c. 77.3 m, which is somewhat lower than the actual floor in L2002, probably due to settling. Large pottery fragments, whole mud bricks and an animal scapula were still stuck in the section, lying on top of the plaster floor, which sealed a thick layer of fill mixed with mud bricks and stone collapse. The collapse descends down, lying on top of the lower occupation phase at an elevation of c. 76.5 m. Unlike the situation in the excavated area, the lower occupation phase was covered here with a thick layer of black ash, attesting to destruction by an intense, apparently localized fire.
Large quantities of restorable pottery vessels were recovered mostly from the upper occupation phase. The lower occupation phase yielded smaller amounts of pottery, mainly small potsherds and no intact or restorable vessels. Other finds included fragments of glass vessels (both phases); large quantities of animal bones (both phases); a couple of cowrie shells (lower phase); a few fragments of basalt and granite grinding stones (both phases); bronze coins (lower phase) and a few small pieces of slag, as well as a small number of corroded iron objects, such as nails and possibly tools (both phases).
All of the coins are associated with the lower occupation phase; although badly corroded they apparently date to the fourth century CE. At this point, the duration of the initial occupation of the fort and whether it was abandoned in the fourth or fifth century CE cannot be determined. It is also unknown whether any earlier floors or occupation phases lie below the lower phase. The fort was reoccupied long after its abandonment. The thick debris layer between the two occupation phases accumulated during this long hiatus. There is no evidence that the upper settlement was a military one, although its character cannot yet be determined. The presence of large numbers of bowls, cooking pots, and storage jars suggest a domestic occupation. The new occupants settled in the ruins of the fort, which was now filled with debris, and reused some of the original walls. Preliminary field reading of the pottery indicates that the upper occupation phase is dated to the seventh century CE, although it could extend somewhat earlier or later, from the late sixth to early eighth century CE.