The earlier remains appear to belong to an Early Roman Nabataean caravan station destroyed in the early second century CE by an earthquake.
Six meters west of the fort, a second earlier structure was excavated. This structure abuts the bedrock shelf on which the fort was constructed (Figs. 4, 5). Sections opened over this structure revealed the remains of at least three rooms and a tabun (F-1; Fig. 6). Parts of two rooms (L401/L402 and L601) were excavated down to bedrock and a small section (L701) was excavated on the opposite side of the building's northern wall (W1). Pottery found in the excavation of the building shows that it was founded in the mid-first century CE and continued to be used until sometime in the early second century CE, when it was evidently destroyed in an earthquake (Fig. 7). Wall 1 appears to have collapsed northward (Fig. 8) and the remains of a cooking pot (Fig. 10:8) in L601, next to the tabun (F-1), had broke and was partially spread eastward next to the interior of W1. Tabun F-1 contained small rocks and a number of potsherds, including an early type of a Gaza wine jar (Fig. 10:11) that is dated to the first–third centuries CE. Other diagnostic potsherds included parts of Nabataean painted ware bowls (Fig. 9:1–8), an Eastern Sigilatta ware bowl (Fig. 10:1), undecorated cups and bowls (Fig. 10:2–5, 7), Nabataean rouletted ware (Fig. 10:6), Nabataean cooking pot (Fig. 10:9), Roman carinated cooking pot (Fig. 10:10), jars (Fig. 10:12, 13), Nabataean strainer jugs (Fig. 10:14, 15) and a fragment of a Roman lamp with a decorated discus (Fig. 10:16).
Visual investigation of the area north of the early structure shows traces of possible wall lines and other rooms. However, no plan of this structure can be determined without carrying out further excavations. It may be assumed, on the basis of the 2004 excavation, that rooms were situated around an open courtyard. The structure was badly damaged by an earthquake and appears to have been stripped of masonry stones nearly to its foundation. Piles of building stones located on the pass below the structure may have been created when the building was robbed. A smaller structure (the Roman fort) was eventually constructed on the bedrock shelf directly east of the early structure.
In addition to excavations in the early building, the exterior sides of the Roman fort along the eastern, northern and part of the western side, were excavated to facilitate restoration work on the structure (L101/L102, L103, L201 and L801). A deep probe along the northeastern corner of the structure (L103) was excavated down to bedrock. In the foundation trench near bedrock a diagnostic fragment of a Late Roman-Nabataean debased painted ware bowl (Fig. 9:9) was found. The structure showed signs of earthquake damage along its northern wall (L201) and the center of this wall had collapsed northward. A section of collapse at this point of the wall was preserved and left unexcavated.
The current excavation confirmed the discovery that the Mezad Mahmal fort is a Roman and not a Nabataean fort, as has generally been assumed. The fort was constructed in the later half of the second century CE in the Late Roman period. It appears to belong to a Roman military initiative of constructing tower forts in the Severan period elsewhere along the Petra–Gaza road, such as the fort of Horbat Qazra and Mezad Neqarot. Other forts of this type and period are known at Horbat Haluqim (‘Atiqot 11 [ES]:34–50) and Horbat Dafit (ESI 3:16–17).
The primary discovery in this season was the remains of the Nabataean caravan station of the first century CE, situated at the head of the pass. This structure, which apparently contained a number of rooms located around a central courtyard, was destroyed in a seismic event in the early second century CE and subsequently, was probably abandoned.