The pool (in excess of 400 sq m; Figs. 2, 3) was carefully built inside a large pit that was dug in the ground. Its walls are not parallel (length of northern wall 24 m, length of southern wall 20.5 m). Five of the original six vaults that covered the pool are preserved today; they are borne on fifteen square pillars and sixteen symmetric cross-shaped pillars, connected by pointed arches. A pointed arch exists between each pair of pillars and a small window, also in the shape of a pointed arch, was installed in the wall built above the arch (Fig. 4). The pillars and arches were built of ashlars, whereas the walls of the pool and the vaults were built of small stones (thickness c. 1 m). A staircase built next to the wall leads from the entrance in the northeastern corner down to the bottom of the pool (Fig. 5). The steps were built of stone; the upper part of the staircase is supported on a flat arch, while the lower part is borne atop another half-arch. Metal steps are placed today on top of the stone steps. Twenty-four square openings for drawing water were originally fixed in the pool’s ceiling. A column base with a round perforation in its center was placed in each of the openings; only twelve have survived today, probably because of rehabilitation work. A square opening was originally installed in the upper part of the pool’s western wall, near the northwestern corner.  

The excavation was conducted during rehabilitation work on the pool, with the aim of determining its construction technique, locating the original water inlet into the pool and ascertaining the pool’s current source of water, which was presumed to be percolating via cracks in the concrete wall that was built by the Mandatory Department of Antiquities. Two half squares were excavated, one inside (Area A) and one outside the pool’s northwestern corner (Area B).
 
Area A. As part of the rehabilitation work, the pool was drained of water and its bottom, which was mostly paved with limestone pebbles, was exposed. The plaster that coated the walls also covered the edge of the floor. At the junction point of the floor and the wall, the plaster was applied diagonally to prevent a right angle where dirt was likely to accumulate. A pavement of marble flagstones in secondary use (L1; Fig. 6) was found in just one spot, under the vault in the northwestern corner, beneath the square opening in the upper part of the western wall. The marble pavement only abutted the western wall of the pool. When this floor section was revealed, it seemed to corroborate the supposition that the water entered the pool via the square opening in the western wall, located c. 7 m high, and the marble floor was designed to withstand the pressure of the water falling from this height. When the bottom of the pool was exposed, a round pit was discovered next to the bottom end of the staircase; three shallow channels dug in the bottom of the pool led to the pit. Today, an electric pump that drains water from the pool on average once every two weeks is located in the pit. It seems that the pit was dug in 1942 for placing a pump inside it. It also became clear that a well-built thickening of stone and impermeable mortar was built around the pillars. This thickening appears in old photographs, but was forgotten over the years.
The northern row of marble flagstones and part of the pebble pavement were dismantled in the excavation of Area A – the northwestern corner of the pool (Fig. 7). It was ascertained that the pool’s foundation was well built (thickness c. 0.65 m) and could only be broken by means of a rotary hammer (electric chisel). A layer of dark gray impermeable mortar (thickness 6–8 mm) that was composed of lime and large quantities of ash was exposed beneath the floor of the pool in the upper part of the foundation. Below the layer of mortar was a layer of potsherds that are difficult to identify, although most of them apparently belong to Gaza jars that are dated to the Byzantine period. The potsherds layer was placed on another layer of impermeable mortar (thickness c. 9 cm) mixed with small stones (L2). Below the layer of mortar and small stones was a thick foundation of stones (L3; thickness 0.5 m), arranged in three superposed courses, with a layer of impermeable mortar (thickness 3–4 cm) between each course. The stone foundation was laid directly on top of pale red sand (L4).
The quality and composition of the thickening built around the columns and pillars resemble those of the pool’s foundation. The base of one of the pillars was inspected and it became clear that a course of ashlars, two stones wide, was built below the pillar (Fig. 8). The thickening was dismantled on the eastern side, where the stones protrude c. 0.15 m from the pillar. No similar protrusion was discerned on the southern side of the pillar. The western and northern sides of the pillar were not examined. Photographs taken of the pool in the past show a pillar whose thickening on the base is damaged, but no ashlar construction that protruded from the width of the pillar is visible below it (Fig. 4). No protruding course of ashlars was discerned in other pillars where the thickening was damaged. The examination of the pillar base also proved that the floor of the pool did not abut the pillar and a narrow space (width 3 cm) was left between them. It seems that for some reason, possibly because limestone pebbles were used, the builders could not join the pebble floor to the pillar and so they left a space between them. The edges of the pebble pavement are raised and the thickening around the base of the pillars was built so as to prevent water from seeping under through the space in the floor.
 
Area B (Figs. 9–11). The excavation was conducted in the presumed location of a channel that supplied the Pool of the Arches with water by way of the high square opening in the western wall. First, a backhoe was used to remove parts of a sidewalk that was built on this section of the pool’s roof. While removing the fill below the sidewalk, the backhoe dislodged two large ashlars (0.3×0.3×0.6 m) that have a hewn groove (thickness 1 cm) at one of their ends on the bottom part. Fragments of a bowl (Fig. 12:6) and a jar (Fig. 12:11) that are dated to the ninth–tenth centuries CE were discovered beneath the sidewalk. Below the sidewalk’s foundation was an accumulation of soil, in which a coin from the Ottoman period (end of the nineteenth century CE) was discovered. A layer of stones and grayish white mortar, which extended across the southern half of the excavation, was exposed below the soil accumulation; the mortar was composed of a large quantity of lime and contained ash and charred olive pits. When this layer was dismantled, a settling pool (L21; Fig. 13) that was delimited by four walls (W101–W104), built of stones, bonded with gray cement and coated with impermeable pale pink plaster, was uncovered. The settling pool was not parallel to the Pool of the Arches and its eastern and western walls are higher than the other walls. When excavating the settling pool, it turned out that the layer of stones and mortar exposed in the square was actually a collapsed vault that originally rested on the northern and southern walls of the settling pool. The negative of the vault’s entire height is clearly evident on W102. An opening of a channel (0.44×0.57 m), coated with impermeable pale pink plaster, was discovered in W102 of the settling pool. Another similarly plastered channel opening (0.30×0.37 m), which conveyed water to the high square opening in the western wall of the Pool of the Arches, was discovered in W101. Both channels were filled with soil. The two openings in the settling pool are not positioned opposite each other and at this stage of the research, it is unclear if this is intentional and what was the reason behind it. It was possibly done to cause the water to flow in a circular motion in the settling pool or this may have stemmed from a constraint associated with the direction of the feeder channel. It also became clear that as a result of this, the settling pool is not parallel to the Pool of the Arches; the channel that connects them turns at some point toward the square opening in the Pool of the Arches. The question arises here too, if this was intentional and what was the reason behind it. The excavation of the settling pool was not completed and its bottom was not exposed. The excavation of the settling pool yielded large fragments of pottery vessels that were discarded inside it, apparently after the pool was no longer in use. These potsherds dated to the ninth and first half of the tenth centuries CE (the time of the Tulunide dynasty) and included bowls (Fig. 12:3–5, 7), kraters (Fig. 12:8, 9), a cooking pot (Fig. 12:10), jars (Fig. 12:12–14) and delicate buff-ware jugs (Fig. 12:15–17). It therefore seems that the settling pool went out of use during the first half of the tenth century CE. This dating matches that of the Gezer aqueduct, when it ceased to be used (HA-ESI 117). It seems that the aqueduct and the Pool of the Arches operated for approximately 150 years and when the aqueduct ceased to be used, the water supply to the Pool of the Arches was suspended as well. 
An accumulation of soil (L22) was excavated north of the settling pool and a stone wall (W100) was exposed below it. The wall, preserved one foundation course high, was founded on soil. It was built in a different direction than the settling pool and it cuts the northern part of the latter. Wall 100 was probably damaged during the rehabilitation work at the Pool of the Arches in 1992, along its northern wall. The soil accumulation contained potsherds, mostly from the ninth century CE, and two from the Umayyad period (eighth century CE), including an Egyptian red-slipped bowl (Fig. 12:1) and a gray jug (Fig. 12:2). It therefore seems that the wall is part of a building that stood there prior to the construction of the Pool of the Arches.
 
The excavation shows that a great deal of effort was invested in the construction of the Pool of the Arches and high-quality materials were used. It seems that a channel, which was connected to the aqueduct from Gezer, conveyed water to the settling pool in antiquity. The cessation of the Aqueducts’ use in the first half of the tenth century CE halted the water supply to the Pool of the Arches, which also went out of use. Following Ramla’s destruction by an earthquake in 1068 CE, the pool remained outside the city limits and was disregarded for centuries.
Concerning the source of water that is seeping into the pool today through cracks in the concrete wall, the idea that it originated, today or in antiquity, from the aquifer or a seep that dictated the location of the pool, is unlikely, considering the historical documentation and the data known today with regard to the depth of the water table. There is no mention of the aqueduct in historical sources from the tenth century CE and these sources state that the city’s water supply relied on cisterns and that its wells were deep and their water was brackish (al-Muqadassi, Ahsah al-taqasim fi marifat al-Aqalim [De Gouja, ed.], Leiden, 1906. P. 164). Therefore, it is unlikely that the pool’s source of water at that time was salty groundwater. According to data provided by the Meqorot Water Company, the water in the Ramla region are currently drawn from an aquifer that is on average 40 meters below ground level, which is considerably deeper than the bottom of the pool. To determine the source of the water today, a sample of the water in the pool and a sample of tap water from the vicinity of the pool were taken. The two samples were sent to a laboratory where it was established that they are almost an identical match. It is therefore clear beyond a doubt that the source of the pool’s water is connected to the municipal water supply. It seems that a local aquifer formed separate from the regional aquifer, into which water seeps from the city pipelines, from irrigating parks, as well as from rainfall. This water can flow on the aquifer from a distance of hundreds of meters from the pool and not necessarily from its immediate vicinity and it is what constitutes the pool’s source of water today.