The site of ‘Ein Naqa‘a was surveyed in the past. Marcus (1993:58–60) described a water system that transported water from the aquifer southward to an open pool via hewn water channels and through a small cave (Figs. 2, 3). The pool was largely carved into the bedrock, but its upper part was stone built. To the east and west of it, Marcus detected multiple side channels of unclear use. At some point after Marcus’s survey, the pool was filled with earth and abandoned. A metal pipe was then laid within a built channel, to carry water southward for about 20 m. In a later survey, Ein Mor described the underground system in detail (Fig. 2), and noted evidence to multiple phases of construction and use (Ein Mor 2013). In 2010, an excavated was conducted at the site by A. Cohen-Klonymous (License No. B-353/2010; Fig. 2).
The current excavation started at the southern end of the water system (north of Cohen-Klonymous’s excavation; Figs. 2, 3) and continued northward along the metal pipe with the aim of locating the pool described by Marcus. Once the pool was identified, the excavation was extended to the west to uncover the well-preserved stone-built channels mentioned by Marcus. Thus, the description below will first discuss the southern channel with the modern pipe and then the northern area comprising the pool and the associated channel system.
 
The Southern Channel
Two parallel walls formed a long channel (W14, W15; Figs. 3, 4) running in a northeast– southwest direction. Set directly upon the bedrock, they were built of medium-sized fieldstones, with some carved blocks in secondary use. Within the channel lay a modern metal pipe, which disturbed the fill of the channel. Hence, it was impossible to securely date the construction of the walls. The fill contained a large quantity of ceramics, most of which dates to the Early Roman period. The channel may have been used as a drainage system for the spring before the modern pipe was installed. The flow rate of the spring, as measured during the excavation at the end of the pipe (Fig. 5), was 1.5 liters/minute.
 
The Northern Area
Excavation in the northern area uncovered the large pool and the system of plastered water channels to its west (Figs. 2, 3). Most of the remains in this area were carved into a large depression in the bedrock and were thus surrounded to the west, north and east by a high bedrock cliff. The cliff-section shows the bedrock to be composed of alternating layers of soft and hard limestone, a structure which is characteristic of the immediate region. When exposed, the soft layers drip water, indicating a high degree of saturation.
The pool is rectangular (5.5 × 7.0 m), and was found completely filled with earth and rubble of large stones (Fig. 6). A small probe in the southwestern corner determined its depth to be c. 2 m, and the fill was then removed using a tractor. As Marcus had noted The lower part of the pool was hewn, whereas its upper part was constructed of fieldstones and coated with a layer of grey cement (Fig. 7). A narrow channel brings water to the northeastern corner of the pool from the underground channel system through a small cave (Fig. 8; previously described by Ein Mor and by Marcus). Although it is impossible to date the construction of the pool, at least some, if not all of its elements may have been contemporaneous with the channels to its west and were regularly maintained until modern times. 
Immediately to the west of the pool there is a system of interlinking channels. Four built and plastered channels, running parallel in an east–west direction, are connected in the east by a slightly wider, perpendicular (north–south) channel. Faint traces of additional stone walls of similar orientation, seen on the surface beyond the excavation boundaries to the south, suggest that additional channels exists in the area. The excavation uncovered evidence of four distinct phases of use and reconfiguration of the channels.
 
Phase I. This phase comprises the initial construction and layout of the channel system (T1–T6). It included an upper section on the bedrock ridge in the north, and a lower section sunk into the bedrock depression.
The upper system was barely excavated, and is therefore not fully understood. It consists of the remains of a C-shaped channel (T6) in the northwestern corner of the excavation. The northernmost section of this channel, which continues east beyond the excavation boundaries, was formed by two parallel walls (W8, W9) that were built directly on the bedrock from small and medium-sized fieldstones (Fig. 9). The inner faces of the walls were coated with white plaster (Fig. 10). Although the southern part of the channel was not preserved, its outline can be reconstructed, showing that it joined Channel T1. This is supported by the contiguity between the bedrock surface of the channel and the southern face of W8 (Fig. 11).
The lower system consists of four built channels (T1–T4), running parallel in an east–west direction and joining in the east a wider, north–south channel (T5). Channels T1–T4 were separated by walls constructed of two rows of large, naturally occurring rectangular stones (W1–W3, W7; average preserved height 1 m). The northern channel (T1), is bounded on the north by the bedrock cliff and on the south by W1 (Fig. 12). It was not completely excavated because the earth fill was saturated with water. In the small portion that was excavated, no remains of plaster were preserved. Flat stone slabs set in a corbelled fashion covered the western part of Channels T2 and T3 (Fig. 13). It is highly likely that all the channels were originally covered in a similar manner, and the undersides of the stones were covered by a layer of plaster. The main, north–south channel (T5), was a little wider than the others and similarly covered with a layer of plaster (Fig. 14). An abundant quantity of ceramic sherds was embedded in the plaster coating of this channel (see below).
 
Phase II. During this phase, three blocking walls (W4, W6, W7) put Channel T5 out of use. Wall 4 adjoined the northeastern corner of W3 and continued to the north in an S-shape curve into Channel T5. The wall was built of a single row of fieldstones, and was preserved to a maximum height of five courses (Fig. 15). Walls 6 and 7, similarly built, bordered Channel T4 and also blocked Channel T5.
 
Phases III and IV. During Phase III, the channels were put out of use and filled with earth and stones. The reason for the seemingly abrupt abandonment of the system is not fully understood. In modern times—Phase IV—the pool was put into use, likely reusing part of the original ancient remains, and was left open. It underwent constant maintenance and was eventually re-plastered with a layer of cement. The maintenance of the pool caused only minimal damage to the channel system in the west.
 
The Ceramic Assemblage
Debora Sandhaus-Re’em and Benyamin Storchan
 
The ceramic assemblage is presented according to the archaeological phases described above.
 
Phase I. The pottery assemblage of Phase I dates to the construction, or the re-plastering, of the water channels. It comprises sherds that were embedded in the plaster lining of Channel T5 (Fig. 16) and includes a bowl with a rilled rim (Fig. 16:2) characteristic of Legionary assemblages, such as the ones from the Convention Center in Binyane Ha-Umma (Magness 2005:69–194) and the Crown Plaza Hotel assemblage (R. Rosenthal-Heginbottom, R. Be’eri, D. Levi and D. Sandhaus, pers. Comm.).
 
Phase II. The ceramic assemblage of Phase II derives from loci around and between the walls that blocked Channel T5. It comprises mostly vessel-types and forms characteristic of early Legionary assemblages (Fig. 17), before the appearance of the arched-rim and rouletted bowls. The types include small bowls (Fig. 17:1–3); large bowls with shelf rims (Fig. 17:4); small casseroles with horizontal ridged handles (Fig. 17:6, 7); cooking pots with an internal groove (Fig. 17:8) and the typical cooking pot with the ridged rim (Fig. 17:9); various types of jugs (Fig. 17:10–12); and the fragment of an amphora (Fig. 17:13). Only a single storage-jar fragment with an infolded rim (Fig. 17:14) can be dated to a later period. This rim is common in assemblages of the third–fourth centuries CE (e.g., Giv‘ati Parking Lot; unpublished). This sherd may date the blockage of Channel T5 to the third century CE. This, however, is unlikely since the overwhelming majority of the ceramics associated with Phase II dates to the first–second centuries CE. Finally, a poorly fired, handmade, deep basin (Fig. 17:5) was probably used for industrial purposes, such as pouring plaster (B. Dolinka, pers. comm.), or it may be a planting pot, a form well attested in the Roman realm (Gleason and Bar-Nathan 2013:317–366).
 
Phase III. The ceramic assemblage from Phase III comprises material from the fill of Channels T1–T3. The pottery is very homogeneous and characteristic of the early Legionary-period assemblages of the mid first–second century CE (Fig. 18). The assemblage includes a large variety of vessel types and forms, including small bowls (Fig. 18:1, 2); large bowls with slightly out-curved rims (Fig. 18:3) or flat shelf rims (Fig. 18:4); basins with grooved ledge rim (Fig. 18:5) or with a very pronounce internal groove in the rim (Fig. 18:6, 7); delicate casseroles with horizontal ridged handles and their counterpart lids (Fig. 18:8, 9); cooking pots of the cauldron type (Fig. 18:10–12); storage jars with a thickened rim (Fig. 18:13, 14) or with an out-folded rim (Fig. 18:15–17); jugs (Fig. 18:18); and a juglet with a cup-shape rim (Fig. 18:19). A single fragment of a roof tile was also found (Fig. 18:20).
 
Phase IV. Pottery of this, latest phase comes from the top soil, representing post-abandonment activity. In general, the pottery is consistent with the earlier phases, and includes mostly Roman-period forms typical of Legion activity (Fig. 19:1–10). Only a few sherds can be dated to a later period, including a fragment of a Late-Islamic-period bowl (Fig. 19:11), a storage jar (Fig. 19:12) and an Ottoman-period tobacco pipe (Fig. 19:13).
The ceramic assemblage from the excavation is quite homogeneous and contains only minimal intrusive material and no rouletted or arched-rim bowls. This strongly supports a date in the mid first–second centuries CE for the water system. The typical Legion-style pottery may also suggest that the system was constructed by the Roman Army.
 
Additional Finds
Three coins were found. Two, a silver para (nineteenth century CE; IAA No. 143886) and a bronze coin of Alexander Jannaeus (80/79 BCE and later; IAA No. 143887), were found by a metal detector before the excavation, and are out of context. The third, a bronze coin of either Herod or his son Archelaus (37 BCE – 6 CE; IAA No. 143888), was found embedded in the plaster at the base of Channel T3 and can be associated with the initial building of the channels (Phase I).
Two metal artifacts were found by a metal detector in the heap of excavated dirt: a small rectangular bronze plaque with rounded corners and a single circular perforation at one end (Fig. 20:1); and a fragment of a bronze rod with circular section and a rounded tip (Fig. 20:2). The plaque is somewhat reminiscent of the scales of Roman equine armor, such as the ones that were found at Gamla (G. Stiebel, pers. comm.). The rod is probably the base of a spatula or kohl stick. The absence of any decoration or stylistically identifiable features on the items prevents further conclusions.
A single carved and incised limestone block reminiscent of a manhole or drain cover (Fig. 21), was uncovered in the fill of Channel T6 (Phase III). Just inside the edge there is a finely incised double line, seemingly part of a circle. A circular hole fully perforates the stone. The stone is too small and fragmentary to identify its function or use.
 
The remains of the water system uncovered at the spring of ‘Ain Naqa‘a were surprisingly well preserved. Both the ceramics finds and the Herodian coin that were found embedded in the plaster of one of the channels date its construction to the first century CE, although some of the underground system may be earlier. The precise way in which the system functioned is not clear, since Channel T6, which apparently carried the water into Channel T1, is at a higher level than the cave through which the spring water flowed, and another source of water is highly unlikely. It is possible that Channel T5 was fed by water runoff through the upper system in the west and received spring water through the cave to the east. The purpose of such an elaborate system is still unclear, but may be related to collection, purification, pressure regulation, and/or distribution of the water.
The site of ‘Ain Naqa‘a is of major importance due to its close proximity to the Emmaus–Jerusalem road. Given its logistic and strategic position, it would undoubtedly have served as a convenient watering stop for travelers en-route to or from Jerusalem.