In the early twentieth century, A. Saarisalo identified some large basalt quadrangular pipe stones lying next to the old road from the Moshava Kinneret to Qevuz
at Kinneret as the Roman underground water pipe to Khirbat al-Kerak/Tel Bet Yerah
. He interpreted a ruin visible at Khirbat al-Kerak as a Roman to Early Arab period fortress, on the basis of potsherds he found around the site (Saarisalo 1927: 15, 76–80
). The archaeological excavations at Tel Bet Yerah
in 1945, uncovered a bathhouse on the tell that was tentatively dated to the fourth and fifth centuries CE (Stekelis and Avi-Yonah 1947: 33–36
). Following the exposure of the bathhouse, Amiran astutely pointed out that a section of the basalt block water pipe, visible in the old riverbed at the western edge of tell, was almost certainly constructed to transport water from the Berenice aqueduct to the bathhouse; the latter and the pipeline must therefore be contemporary (Amiran 1955: 39–46
). Many of these basalt pipe blocks that must have been exposed in the course of laying water and irrigation pipes over the past century are exhibited until today, individually or in groups, in the courtyards of neighboring settlements.
However, the specific location of the pipe was not recorded and the current excavation is the first time that the basalt pipe was excavated in situ, and an underlying terracotta pipe was discovered. It is evident that the two pipelines were built to divert water off from the Berenice Aqueduct, which channeled water from springs in Wadi Fejjas near Yavne’el to the city of Tiberias, and to transport it downhill across the Jordan riverbed and uphill again to Tel Bet Yerah, using the inverted siphon technology.
The two pipelines ran downhill from west to east, following exactly the same line; the bottom of the basalt pipe blocks overlaid the top of the terracotta pipe segments by c. 0.6 m, indicating that the terracotta pipeline clearly predated the basalt pipeline, at least technically (Fig. 3).
The pipelines are described below and their origin and destination are examined so that they can be understood in their proper context.
The Terracotta Pipeline. At a depth of c. 1.3 m below the surface, two separate short stretches of a terracotta pipeline (L103—segment length 3.5 m, L109—segment length 1.2 m) were exposed where the overlying basalt pipeline blocks had previously been removed by modern earthworks. At the end of the excavation, when the newly-found basalt blocks were removed under IAA supervision, to allow the laying of the modern pipes, an additional 8.3 m long section of the terracotta pipeline, lying between the two shorter excavated sections and slightly farther to the west, was uncovered with a backhoe. The total exposure of the terracotta pipeline was nearly 13 m, consisting of 55 articulated pipe segments; the section examined in detail (L103; length 3.5 m) consisted of thirteen joined cylindrical segments. The east-facing end of each segment had a cylindrical projection with a reduced diameter that was inserted into the wider and slightly out-flaring west-facing end of the adjacent segment (Fig. 4). The shape of the pipe segments was basically the same, but sufficient differences indicate that they were handmade, rather than cast in molds. Their dimensions were rather similar (average length including cylindrical projection 22 cm; length of eastern projection 5 cm; outer pipe diam. 17 cm; thickness of wall 1.9–2.2 cm). The slightly flared-out form of the pipe segments and the method of attachment resulted in an uneven internal diameter (mid segment inner diam. 9–10 cm, inner diam. at joins, c. 8 cm; Fig. 5). Careful preparations were made on the ground prior to laying the pipe. A small square-profiled trench (width 0.4 m, height 0.5 m) was cut into the natural clay-like alluvium and a layer of a light-colored, aerated mud-brick-like material (thickness 12 cm) was laid on its bottom. It served as a foundation for assembling the pipe segments, whose joins were sealed hermetically with a white lime substance (Fig. 6). The spaces in the sides of the trench and on top of the pipe were then filled with a layer of the same light-colored material (thickness c. 10 cm), so that the terracotta pipe was completely encased and protected at the time of use (see Fig. 3: Sections 2-2, 5-5). The inside of the pipe was examined by inserting an arm through a broken pipe segment; it revealed that the pipe was empty and only a thin layer of travertine had accumulated on the inner surface, probably indicating a very brief period of use. A few small and worn potsherds found in the covering mud-brick layer were mostly non-diagnostic, but included a small Hellenistic jar rim, a few fragments of Roman fabric and a small Early Islamic buff fabric sherd. Samples of the various materials used for the laying of the pipe were collected for future examination.
The Basalt Pipeline. The large basalt pipe was laid c. 0.5 m above the mud-brick layer that encased the terracotta pipeline, by people who were familiar with the exact line of the earlier terracotta pipe (see Fig. 3: Sections 2-2, 5-5). Fourteen basalt blocks were uncovered in four clusters over a distance of 20 m; the intervening gaps are the result of damage caused by earlier pipe-laying earthworks, twice in the 1960s–1970s, and thirdly in the present pipeline earthworks, when fourteen other blocks were uprooted. At the western end of the excavation, three joined blocks emerged at a depth of c. 1.5 m below the earth slope (L101), the surface here being a raised bank supporting a dirt track. There followed a gap of c. 4.5 m, then three slightly disjointed blocks (L102), a gap of c. 7 m, a length of six joined blocks (L104, L107), a gap of 1.5 m, and finally two blocks at the eastern end of the excavation.
The basalt blocks had several features that were essential for the proper functioning of the pipe (Fig. 7). They were mostly quadrangular or cubic-shaped (length 0.55–0.65 m, width 0.65–0.75 m, height 0.6–0.7 m). The bases were cut fairly smooth to enable the blocks to stand firm; the adjoining sides of consecutive blocks were evenly dressed and the circular horizontal bore (diam. 0.3 m) was uniformly smoothed. The east-facing end of the blocks was carved with a projecting collar (length c. 6 cm), surrounding the bored hole (diam. 0.3 m); the west-facing end had a matching rebated socket, permitting a fairly exact fit. The outer non-contacting block surfaces were roughly worked. Some of the blocks had also a smaller vertical hole (diam. 10–12 cm) bored through the top surface into the pipe; in one block, work on hewing out the vertical hole had begun, but was not finished and the stone was not pierced (Figs. 8, 9). There was no extant evidence regarding how these small ‘vent-holes’ were hermetically sealed to allow the functioning of the pipes under pressure and probably to release pressure when necessary; their purpose was probably also to provide access to clean inside the pipes, Some blocks had other different features, including a cylindrical block with a smoothed outer surface, a block that was a reworked worn Corinthian capital (Fig. 10), and a short block (length 0.3 m) that had a disproportionate long projecting ring (length 0.26 m). In all these blocks the inner diameter of the pipe was constant (0.3 m). As with the clay pipeline beneath it, the massive basalt stone pipeline was laid on a carefully-prepared surface (L105, L108), attained by cutting out a large trench (width c. 2 m) and lining it with some smoothed, slightly concave layers (thickness c. 15 cm) of light grayish-colored earth. The joins between the pipe blocks were hermetically sealed with thick white lime substance and the spaces between the pipe walls and the sides of the trench were filled with small to medium-sized, water-smoothed stones, probably taken from the riverbed or around the lake, whose function was to firmly wedge-in the pipe and prevent it from moving (Fig. 11). Only a few of these small stones overlay the pipe and it is not clear whether the pipe was covered over or not. A small amount of travertine accumulated inside some of the pipe blocks and a few small non-diagnostic body sherds of probably Roman date were found scattered around. A single bronze coin, important for dating, was discovered in-situ adjacent to one of the basalt blocks.
Samples of the various materials used for the laying of the pipe were also collected here for future examination.
The Pottery. As noted above, the potsherds were small, worn and non-diagnostic, but a few Roman body sherds, a Hellenistic storage-jar rim and an Early Islamic buff-ware sherd were identified. However, they are inadequate for dating either the construction of the pipes or their period of use.
. A couple of small glass fragments, found between the small stones around the basalt blocks (L107), were identified as fragments of a light bluish green bottle with applied turquoise threads, characteristic of the Umayyad period; many examples of such bottles have been found at Bet She’an (Hadad 2005: 24, Plate 13
Two bronze coins were recovered from the excavation.
The earliest coin is a surface find, a worn FEL TEMP REPARATIO coin minted in Constantinople and dated to 351–361 CE (IAA 140983).
The second is an Arab-Byzantine copper, related to the Imperial Image series, dating roughly to 660–680 CE (IAA 140982). The coin depicts on its obverse an imperial figure seated on a throne; on the reverse a cursive m flanked by palm fronds, with pellets between the legs of the latter, another pellet below and an emphasized exergue line (Fig. 12). According to typology and style, this coin belongs to a group classified as the Pseudo-Damascus mint, due to its resemblance to those coins struck during the same period and minted in Damascus (Album and Goodwin 2002: Pl. 39, No. 578).
The Origin and Destination of the Pipelines. Since the excavated pipelines were part of an inverted siphon conduit that received water from the Berenice aqueduct, and transported it across the Jordan River to its destination at Tel Bet Yerah, it is necessary to summarize the relevant data on the Berenice aqueduct and Tel Bet Yerah.
The Berenice Aqueduct to Tiberias
. This aqueduct (length c. 15 km) was a stone-built open water channel, originally roofed with stone slabs. It was built to transport fresh water from the sea level springs of 'Ein Fejjas (Arab name of the Greek Pegae
) in Nah
al Yavne’el northward to Tiberias, the city founded by Herod Antipas around 19 CE, where the water was stored in and distributed from a municipal pool located at c. –170 m below sea level. On the basis of a comprehensive survey of the water channel, the construction of the aqueduct was attributed to the Late Roman period, around the fourth century CE (Vinogradov 1988: 151–165
). The channel became popularly known as the Berenice aqueduct; this name was first recorded by nineteenth century travelers after the sister of Agrippa II who lived in the late first century CE, probably since the channel ran northward along the mountain slope of the same name, on which a Byzantine church was later built. The remains of a smaller stone-built open channel were subsequently observed underlying this channel, at a mid-way spot along the channel route, and it was suggested that the earlier channel may date to the first century CE (Tsuk 2011:124–135
). The Berenice aqueduct entered the city of Tiberias from the south; in the recent excavations of the Roman theatre, it was observed that the remains of the Abbasid eighth century CE domestic neighborhood that was built over the theatre, cut through the water channel, indicating that it was no longer in use by this time (W. Atrash, pers. comm.). It thus seems that the Berenice aqueduct's maximum lifespan was from the Early Roman to the Umayyad periods (first or second to early eighth centuries CE). The aqueduct ran northward on mid-slope, directly above the extrapolated line of the inverted siphon pipe and was undoubtedly the water source for the pipelines.
Tel Bet Yerah
(Khirbat al-Karak). Several archaeological excavations carried out since the mid-twentieth century have revealed that the site was settled on a massive scale in the Early Bronze Age, and was rebuilt as a polis
named Philoteria in the Early Hellenistic period. A fortified complex excavated in the 1950s was identified as a Roman-Byzantine-period synagogue, on the basis of a column base engraved with a seven-branched menorah; however, this interpretation was subsequently refuted (Reich 1993
: 137–144;HA-ESI 122
), as was the Roman-Byzantine date proposed for the adjacent bathhouse.
The fortified complex, incorporating the adjoining elaborate bathhouse have since been reinterpreted as the Early Islamic qasr
of al-Sinnabra, the winter palace constructed and occupied by the Umayyad Mu‘awiya and Abdel Malik between 640–705 CE (Whitcomb 2009: 241–242
). Additional archaeological evidence from the current excavations of Tel Aviv University, carried out beneath the bathhouse, has consolidated this interpretation and dating (S. Paz and T. Deadle, pers. comm.). The water for the bathhouse was supplied in still-extant long pipes composed of terracotta segments, leading into the site from the southwest. A petrographic analysis, carried out in the course of the present excavation, of the terracotta pipes from the Early Islamic bathhouse at Tel Bet Yerah
and the terracotta pipes from the newly-excavated pipeline, revealed that the two samples share the same petrographic affinities and the same local provenance. The Early Islamic al-Sinnabra palace and bathhouse were thus the destination of the pipelines.
The Reconstructed Course of the Inverted Siphon Pipelines.
Vitruvius, the first century BCE Roman architect and engineer describes the technology of inverted siphon clay and lead pipelines that were employed to overcome topographical obstacles in long-distance water systems (Vitruvius, Vol. VIII: Chapter VI
). Although stone masonry was standard in Vitruvius’ day for building open water channels, he does not mention the use of stone for pressure pipes, and it is possible that stone pipe blocksfor inverted siphon pipelines were first implemented in the course of the Early Roman period.
A possible reconstruction of the siphon pipelines’ course is presented here. The line of the Berenice aqueduct ran along the hill slope above the pipes exposed in our excavation, at a level of c. –150 m below sea level, whereas the elevation of the westernmost highest basalt block, uncovered c. 600 m farther to the east, was c. 50 m lower at –200.6 m below sea level (Fig. 13). The pipe had to descend to the old Jordan riverbed at an elevation of –210 m below sea level and to rise again to Tel Bet Yerah at an elevation of –190 m below sea level; for the inverted siphon pipe to function, it only had to begin at a level that was slightly higher than the end point, possibly at an elevation of –180 m below sea level. A small difference between the levels of the beginning and the end of the closed pipe was preferable as there would be less pressure inside the pipe. It is therefore probable that there was an additional open channel leading off from the Berenice channel in a northeasterly direction down to a header water tank located on the hill at an elevation of c. –180 m below sea level, in which case the entire inverted siphon pipe might have been c. 800–900 m long. This header tank, which was essential for regulating the water entering the pipe and for trapping the silt, may be looked for in the future somewhere below the line of the aqueduct and above the pipe. The pipelines had to cross the Jordan riverbed, which required a short bridge on which the sealed pipe would be laid above ground, raising the lowest point, possibly by c. 3 m to –207 m. Vitruvius points out that the lowest part of the inverted siphon pipe should be a long level stretch, rather than an elbow-shaped bend, which was far more likely to burst the joints of the pipes. The earlier terracotta pipe may have been problematic on two levels: firstly, it had a small inner diameter (8–9 cm), which could only supply a limited amount of water and secondly, the pressure may have caused the thin-walled pipe to burst, presumably at its lowest point at the Jordan crossing, where the pipe lay above ground. The basalt pipe that replaced the terracotta one was far stronger and had a much larger capacity. These pipelines reached the western side of Tel Bet Yerah, where there was probably a water tank to store, regulate and distribute the water; the terracotta pipe supplying water to the bathhouse would have led off from this distribution tank.
The Dating of the Inverted Siphon Pipelines: The archaeological data for dating the inverted siphon pipelines includes the single in situ coin and the glass fragment, as well as the contextual data from the Berenice aqueduct and the al-Sinnabra palace and bathhouse. All the data indicate an Umayyad date (mid seventh or early eighth centuries CE) for the pipelines. It is possible that future chemical analyses of the lime and mortar materials employed in both pipelines will provide additional data.
The Origin of the Basalt Blocks and the Hippos-Sussita Inverted Siphon Pipeline:
The issue still to be solved concerns the large basalt blocks and whether they were purposefully cut for the Umayyad pipeline. This question was raised in the 1990s when the water-supply system to Hippos-Sussita, the Hellenistic–Roman Decapolis city on the east side of the Sea of Galilee, was examined and the similarity of the stray Tel Bet Yerah
blocks to the Sussita blocks was observed. Spring waters were supplied to Hippos in long-distance water channels that had to traverse a 40 m deep saddle southeast of the city—a topographical obstacle that necessitated the employment of a 600 m long inverted siphon pressure pipe. A thorough survey of the water-supply system to Hippos that looked for the inverted siphon pipeline uncovered only fragments of basalt blocks along the saddle, as well as a fragment of a single block in-situ
(Meshel, Tsuk, Fahlbusch and Peleg 1998: 33–39, 72–73
). The western continuation of the inverted siphon basalt pipeline, after it leveled out, was uncovered during the Sussita excavations, running under the decumanus maximus
street from the eastern city gate into the city. The construction of the Hippos water system was attributed to the Roman period, probably the first century CE, and it may have functioned until Sussita was finally destroyed in the earthquake of 749 CE (Segal 2008: 1782–1787
). However, it seems that the city experienced some decline after the Arab conquest of Palestine in the mid-seventh century CE, and not all the churches continued to function into the eighth century CE (M. Eisenberg, pers. comm.). In the course of the survey, Y. Peleg pointed out the similar appearance and dimensions of the stray basalt blocks from the area west of Tel Bet Yerah
to the blocks from Hippos. A petrographic analysis carried out on the basalt blocks from both sites was inconclusive and Peleg cautiously suggested that the Bet Yerah
blocks may have originated from Hippos and have been employed in secondary use in the inverted siphon pipeline to Tel Bet Yerah
, thus explaining their disappearance from the former site. This could only have been possible if the Sussita pipeline was dismantled when the Umayyad al-Sennabra palace and bathhouse functioned around the late seventh or early eighth centuries CE, thus leaving the eighth-century CE Christian population of Sussita entirely dependent on water collected in the many extant cisterns. The rather strange juxtaposition of various styles of basalt blocks that was uncovered in our excavation supports the interpretation that the blocks were in secondary use in the Tel Bet Yerah
siphon; the cylindrical block and the narrow block with the long projection probably had specific functions in their original Hippos context. The transportation of the blocks by floating them from the eastern side of the Lake of Galilee to its western side and then along the ancient Jordan riverbed would have been feasible, and certainly involved a lesser effort than preparing hundreds of new blocks. If this scenario is correct, the Umayyads emulated the Roman system of supplying water in basalt pipes, using the inverted siphon technology, by dismantling the Roman inverted siphon at Hippos-Sussita and reassembling it at Tel Bet Yerah
On the basis of the combined evidence, it is proposed that the terracotta pipeline and the overlying basalt pipeline were both constructed in the Early Islamic period to transport water via pressure pipelines from a junction in the Berenice channel to the Umayyad palace and bathhouse at Tel Bet Yerah. It is probable that the basalt block pipeline was constructed to replace the original clay pipeline that did not function effectively or may not have withstood the pressure. It is feasible that the large basalt blocks of the Roman inverted siphon pipeline, supplying water to Sussita, were dismantled by the Ummayads, floated over the lakeand reassembled into a new pipe at Tel Bet Yerah. The possibility that the clay and basalt pipes were in use contemporaneously but had different functions, seems less likely, but cannot be discounted.