Between Gaulanitis and Hippos: The Roman Road in the Southern Golan Heights in Context

Abstract

The Roman road that transverses the southern Golan is one of the best-preserved ancient roads in the region. The present study provides new evidence regarding its route, its date of construction and abandonment, and its military-administrative and geospatial relationship with the surrounding landscape and settlements. The research combines diverse methodologies, including historical cartographic data, survey, excavation, a mobility model of the region and visibility predictive analysis for watchtower positions. The results suggest that the road was constructed in the second half of the 2nd century CE. A system of watchtowers existed along it, and it apparently did not serve as a vital artery between the nearby villages and the road itself.

Keywords:

• Roman roads
• Roman watchtowers
• Archaeological GIS
• Roman administration
• Spatial analysis
• Hippos territorium

Background and previous research

Roman roads have been documented in the northern, central and southern Golan. The Roman Imperial highway that crosses the southern and central Golan Heights (Fig. 1) is one of the best-preserved sections in the region. The region under study extends roughly within the Hippos territorium—between the Sea of Galilee in the west, the Ruqqad and Yarmouk Rivers to the east and south, and south of the Daliyyot gorge to the north (Fig. 2a).¹ The region is defined by a basalt plateau sloping from northeast to southwest and dropping sharply to the Jordan Rift Valley. It is shaped by Cover Basalt and Pleistocene Dalwe Basalt formations in the northeast. Deep and fertile soils developed in the south, whereas the soils in the northeast are shallower and rockier. The plateau is dissected by deep valleys that run in east–west and north–south directions, emptying into the Jordan Rift Valley or the Ruqqad and Yarmouk Rivers and exposing sandstone, limestone and chalk formations (Mor et al. 1997).

Fig. 1: The Roman road system in the Galilee and southern Syria; the roads under study are marked with a dashed line (prepared by I. Ben-Ezra based on a map prepared by A. Pazout)

Fig. 2: a) The study region with principal sites mentioned in the text; b) the ‘natural corridors of movement’ (mobility model) based on CFMN analysis; the colour scale indicates cumulative line density (lines per km²) for the computed focal mobility networks classified as 1/2 standard deviation; higher numbers indicate where computed movement corridors converge

Segments of this road have been previously described—albeit without detailed discussion—as part of the Roman road network in the Golan. The observations and results of the previous surveys are included below.² Twelve milestones in all were found in four areas. A watchtower along the road was excavated in 1981 by Maʿoz, and the construction was dated to the end of the 2nd–beginning of the 3rd centuries CE. The primary function of the Roman road network was related to military transportation, the safety of passersby and their merchandise, the connection of military installations with supply centres and administrative hubs (usually major provincial cities) and state postal service (Chevallier 1997: 274–279, 306–308; Roll 2009: 9*–11*).

Research goals and methodology

A new study of the road was recently conducted in order to address research topics that have either not been dealt with or dealt with only in brief. A main goal was to better determine the time of construction and abandonment of this road. In addition, a more precise mapping of the entire route from the Sea of Galilee until the modern-day Israel–Syria border was conducted, including description and measurements of the road in various sections. This was also important in order to better determine the western section of the road between the Sea of Galilee and the Lawiye plateau, which has not been preserved. Furthermore, the landscape setting of the road and in particular its relation to other features (especially the ancient settlements and field systems) was studied for the first time.

In order to address the topics above, a complex approach combining diverse methodologies from GIS-based spatial modelling and a GIS mobility model of the region, as well as visibility analysis for prediction of watchtower positions, field survey and excavation, was adopted. Our methodology included:

1. Compilation and analysis of modern and historical cartographic data pertaining to the road system in the research area and archaeological data pertaining to the road and related installations and ancient settlements;

2. GIS spatial analysis for delimiting areas of special attention, where additional watchtowers/milestations could be expected. These were defined as an area of 150 m radius at increments of 1,500 m (ca. one Roman mile)³ measured from the two known watchtowers/milestations (Hartal and Ben Efraim 2012a: Site No. 79; 2012b: Site No. 111). Furthermore, visibility analysis was performed to define areas visible from the two known sites, given the assumed original height of the towers (ca. 6 m), in order to refine the survey area for locating presumed neighbours (Pažout 2017a);

3. Using Cumulative Focal Mobility Network (CFMN) analysis to identify the route of the road from the Lawiye spur to the Sea of Galilee and its relation to expected movement patterns in the region (Pažout 2017b),⁴ in an effort to examine if the route of descent from the Golan plateau would be the same as the ascent. The focal mobility networks for the source points were extracted using hydrology tools, after computing cost surfaces for each point. Arbitrary friction surface was used, excluding very steep slopes (>40°) and bottoms of valleys (due to seasonal flooding and the ubiquitous presence of waterfall steps) from the analysis. In this case study we are following similar applications of GIS methodologies for prediction of routes of Roman roads (Abou Diwan and Doumit 2017; Fonte, Parcero-Oubiña and Costa-García 2017; Parcero-Oubiña et al. 2019);

4. Field survey and documentation of the visible segments of the Roman road, focusing on its physical description, construction methods, materials and landscape context;

5. Excavations of the Roman watchtower at el-Qusayyibe East⁵ were performed in order to complement data provided by Maʿoz (1982) during his excavations of a neighbouring watchtower to the east and in order to refine dating of the Roman military presence in the area.

The road segments and construction techniques

The general route of the road was described in previous research (see above, n. 1). The westernmost section was not preserved, and the route has been suggested on the basis of the location of milestones (see below). Our analysis, using the methods described above, provides a more detailed description of the various road segments and their construction techniques.

The road crossing the southern and central Golan runs roughly east–west and is composed of two branches (R1 and R2; Fig. 2a). Remains of the pavement (R1) appear on the Lawiye spur between two modern orchards for about 1,400 m. The road reappears east of the orchards; then it can be followed in a straight line for ca. 1,450 m, where it takes an angular turn due south, continues for another ca. 400 m, takes another angular turn due north and pursues a straight line again for ca. 2,000 m, where the road branches. The northern branch continues in an east-northeasterly direction for ca. 4.8 km to Jurniyye. On the basis of the topographical maps and satellite images it is possible to trace the northern branch for an additional ca. 13 km eastward to Nawa in southern Syria, where it apparently joins a road running from Damascus southward (Bauzou 2003: 303–306). Additional branching is indicated on the maps east of Jurniyye, but the road’s course was not surveyed since it runs through a restricted border area.

The southern branch (R2) continues in a straight line due southeast for ca. 5.2 km, where it turns slightly north and continues for ca. 5.4 km to the Roman bridge spanning the Ruqqad River (Jisr er-Ruqqad; Schumacher 1888: 165–167) and further across Nahr ʾAllan (Schumacher 1889: 76–79). Its eastern terminus is not exactly known, but it probably joins the same road from Damascus as the northern branch, albeit south of Nawa.

The western terminus of R1 was not known as no remains of the pavement were found between the lake and the central part of the Lawiye spur. The CFMN model (Fig. 2b; Pažout 2017b) suggests that the road probably proceeds south of Shuqayyif on the western piedmont of Lawiye spur westward and reaching the Sea of Galilee approximately in the area between Kafr ʿAqabiya and the lake shore on the lower Naḥal Kanaf. Here it connects to the road from Caesarea Paneas to Nysa-Scythopolis. The find spots of milestones IMC 212 and 250 on the northwestern terrace of the Lawiye spur conform to our model (Staab, Pažout and Eisenberg 2020). Note however, how the actual route of the road diverts from the CFMN model on the plateau once it leaves the piedmont and ascent, as it deviates from the computed natural movement corridor. A possible explanation for the road not following the best natural route as defined in the model that its course on the plateau was influenced by the intended destination (likely Nawa in southern Syria, beyond our research area), rather than by the topography or geomorphology (slope).

Along the length of the road, all segments are defined by two curbs, a central line of stones (spina) dividing the road into two lanes and a pavement of small stones. Curbs and spina are built of larger, roughly square, basalt fieldstones (up to 0.4 × 0.6 m; Fig. 3a), whereas the pavement is made of smaller fieldstones (diameter 0.1–0.3 m; Fig. 3b). In certain low-lying areas, with risk of flooding during winter, the road is elevated on a stone embankment (such as the crossing of upper Naḥal Samakh and west of Jurniyye). A built ford was earlier reported on the southern branch crossing upper Naḥal Nov (Hartal and Ben Efraim 2012b: Site No. 114). In many places the pavement and curbs are disturbed and only a spina indicates the course of the road (Fig. 3b).

Fig. 3: An example of a typical a) spina and b) curb; c) remains of R2; note the field walls overlying the curbs on the left (photos by A. Pazout)

The current study shows that the two branches differ slightly in the total width measured from the outer edges of the curbs. R1 has a general width of 4.8–5.2 m (ca. 16–17.5 Roman feet), whereas R2 is slightly narrower, with a width of 4.4–4.8 m (ca. 15–16 Roman feet). This may suggest that different workgroups were responsible for the construction of different segments of the same road. Like the variation in the milestone dates in the eastern and western segments (see below), this variation in width may also suggest that the sections were not built at the same time.

Milestones

In total, 12 Roman milestones were identified along the road. Two of them (IMC 464 and 468), located east of the branching, disappeared several years ago,⁶ and no inscriptions are recorded for them. Eight milestones were found in situ next to the watchtowers, but no inscription was identified on them either. Inscriptions on two milestones found on the northwestern slope of Lawiye were read for the first time within the framework of this research; they are dated to the reign of Trebonianus Gallus (251–253 CE) and probably the first Tetrarchy (293–305 CE; Staab, Pažout and Eisenberg 2020). A group of six milestones was found ca. 6 km west of Nawa, southern Syria, on a continuation of the road from Jurniyye eastward. These milestones (Sartre-Fauriat and Sartre 2016: 411–415) date from the reign of Marcus Aurelius and Lucius Verus (161/2 CE) to Septimius Severus (197 CE), thus complementing the picture provided by the Golan milestones. In summary, the earliest date of 161/2 CE from Nawa may provide the terminus post quem for the section that crosses the Golan.

Roman watchtowers and the excavation at el-Qusayyibe East

Two watchtowers along R1 have now been excavated. The watchtower excavated by Maʿoz (1982), dated to the end of the 2nd–beginning of the 3rd century, was noted above. The watchtower at el-Qusayyibe East (Hartal and Ben Efraim 2012a: Site No. 79), excavated within the framework of the current study, is located ca. 16 m north of the Roman road and west of the previously mentioned tower. Four milestones or their fragments, without inscriptions, were previously surveyed in the vicinity (IMC 293, 408, 409 and 415; Fig. 4). The walls of the small rectangular tower (ca. 5.9 × 5.55 m) have survived up to two courses high; they are made of dry masonry (Fig. 5). They are built with two faces of hewn basalt ashlars laid as headers-and-stretchers or as headers. The foundation course is composed of long (ca. 0.8 m) headers. The walls are between 0.85–0.9 m in width. The western wall (W161) is mostly conjectural, as the excavations in the western part of the structure did not continue below topsoil. Two segments of a possible partition wall (W166) were uncovered adjacent to the northern and southern walls, creating two rooms (ca. 4.2 × 1.55 m each). The wall (ca. 0.8 m wide) is built with two faces of basalt fieldstones. Since the central part of the wall was not uncovered (or seen in the section,) it is possible that the two segments are a later addition of pilasters/buttresses supporting the ceiling, or perhaps an arch, although no remains of vaulting were identified. The 0.8 m wide entrance lies in the centre of the eastern wall (W159), with a threshold in situ. A basalt door was found when clearing the topsoil above the northeastern corner. As it is too small for the threshold, it may have been used to shutter only part of it.

Fig. 4: The el-Qusayyibe East watchtower (upper left) in relation to the road and milestones (photo by M. Eisenberg)

Fig. 5: The el-Qusayyibe East watchtower following the excavations (photo by M. Eisenberg)

Excavations in the eastern room of the watchtower uncovered a stone pavement (F163) in the southern half of the probe, apparently composed of small basalt stones set in a dark clay soil. Locus 165, below F163, was composed of more loose dark brown soil with less inclusion of basalt rocks. Given that the inner faces of the walls of the structure at this depth were built of smaller unhewn basalt stones, L165 probably represents the foundation level of the structure.

Pottery dated mainly to the Middle and Late Roman periods was found, the latest dating to the 3rd–4th centuries CE. The assemblage is composed mainly of local wares, such as cooking pots and bowls (Kefar Hananya forms) and storage vessels. A small quantity of Hawarit ware was also found. These are rare at neighbouring settlement sites, but are common in the northern Golan Heights in the 2nd–5th centuries CE and are found in the central Golan mainly in the 4th century CE. Two fragments of a Roman-period discus lamp were also found. No coins were recovered, and the only ancient metal find is an iron hobnail from a Roman military boot (caliga). No clear Byzantine types or imported LRRW are present, although these are common by the late 4th century. The latest pottery assemblage can be dated to the second half of the 3rd–first half of the 4th century CE, providing us with a terminus ante quem.

The watchtower can be compared in plan to the one excavated by Maʿoz (1982). The latter was a square structure (ca. 5.9 × 5.9 m), with walls built in header-and-stretcher masonry and with a wider foundation course of headers. Its construction was dated to the 2nd–3rd centuries CE, on the basis of pottery and coins of Commodus (180–192 CE) and Elagabalus (219–222 CE), but the material from Maʿoz’s excavations was never published in full. Given the evidence from el-Qusayyibe East, we may propose a late 2nd/3rd-century date for the construction of the watchtowers and a date in the first half of the 4th century at the latest for their abandonment. They were clearly constructed after the Early Roman period and were no longer in use in the Byzantine period.

Other examples from the region include watchtowers surveyed along the segment of Via Nova between Bostra and Philadelphia-ʿAmman. These are somewhat smaller (ca. 5 × 5–7 m), with 0.7–0.8 m wide walls built as headers or headers-and-stretchers (Bauzou 1998: 135–139). Watchtowers surveyed along the Leja road are of similar dimensions, but with wider walls (1.2 m) built in the same manner (Bauzou 2003: 297–299). The Leja road was constructed, according to the milestone evidence, between the reign of Marcus Aurelius and Lucius Verus and that of Commodus (161/2–186/7 CE; ibid.: 300–301).

Sur vey of Roman watchtowers

We tried to identify additional watchtowers and mile stations by surveying areas at ca. 1,500 m (one Roman mile) intervals to the west and east of the known watchtowers (Fig. 6). The visibility analysis was designed to refine survey areas of the immediate neighbours of the two known sites—Maʿoz’s watchtower and el-Qusayyibe East (Fig. 7). We assumed that watchtowers were constructed with intervisibility in mind in order to facilitate the relaying of messages through optical signals (Woolliscroft 2001). However, modern activity in some of the areas of research prevented a full investigation.

Fig. 6: Surveyed and excavated watchtowers with the survey areas of the milestations

Fig. 7: Visibility maps; a) Maʿoz’s watchtower; b) el-Qusayyibe East; the height of the observer (watchtower) was set at 5 m above ground, the height of observed objects (OFFSETB) was calculated at 1 m above ground (general visibility of the landscape) and at 5 m above ground (height of the assumed observed neighbouring watchtower)

Remains of a stone structure were observed at a distace ca. 145 m short of four Roman miles from Maʿoz’s watchtower on the northern branch of the road. The structure was located on an elevated basalt flow of Dalwe formation, ca. 7 m north of the road (Fig. 9a). The elevated position was probably chosen for better visibility as the four-mile mark is in a small depression formed by a seasonal stream. Remains of three perpendicular walls, forming an approximately square structure (ca. 4–5 × 5–6 m), are visible. The most visible eastern wall is ca. 0.7–0.8 m wide, built with two faces from roughly hewn basalt blocks (Fig. 9b). A small quantity of Middle/Late Roman pottery was observed on the surface. The site is located ca. 1.6 km west of Jurniyye (i.e., a distance of slightly more than one Roman mile). The structural remains, pottery finds and position along the road leads us to identify this site (named Height Spot 500/Ḥorvat Elah) as an additional Roman watchtower on the northern branch of the road (R1). Thus, in addition to the two previously known sites, only one good candidate for an additional site was identified in the survey. Two sites previously suggested as Roman watchtowers (Hartal and Ben Efraim 2012b: Site Nos. 103 and 116) were resurveyed. Contrary to previous assumptions, no structural remains were identified at either site; moreover, the visibility analysis does not suggest that either site was intervisible with known watchtowers (Fig. 8).

Fig. 8: Location of Site No. 103 in relation to areas visible from Height Spot 500 (Horvat Elah) and Maʿoz’s watchtower

Fig. 9: a) General view of Height Spot 500 (Horvat ʾElah); b) segment of an eastern wall (photos by A. Pazout)

The landscape of the Roman road

The route of the imperial road is surrounded by field plots (rectilinear, often oblong, plots of land defined by dry masonry walls—field walls—of local basalt), stone mounds and some villages. The relationship of the road to the landscape was addressed only in brief in previous research (Urman 1985: 112–116).

The field walls marked on Fig. 10a were digitised in a high-resolution orthophoto⁷ and verified in the field survey. Irregular and oval enclosures are probably animal pens or garden plots, associated with modern settlements and inhabitants of the Golan. The typology and chronology of these structures is not sufficiently known. The rectilinear field plots are best visible south of Mazraʿat Quneitra and around Jurniyye (Fig. 10a); where adjacent to the road, they mostly respect its alignment. In many instances, especially along the northern branch west of Jurniyye, the field walls overlie the curbs and the sides of the road (Fig. 10b). This suggests that these field walls were built after construction of the road and most likely when the road was still functioning (at least as a local access road to the fields). The field systems around Mazraʿat Quneitra were dated to between the 2nd/4th and the 5th/6th centuries CE (Gibson and Lewis 2023: 221); this corresponds to the initial construction and subsequent use of the road.

Fig. 10: a) Field systems that appear to be aligned to the road in the vicinity of Jurniyye (map courtesy of the Survey of Israel); b) one of the field walls around Jurniyye aligned with and overlying the curbs of the road (photo by A. Pazout)

The relationship of the road to the surrounding villages is interesting as there is no large urban centre in the central Golan connected to it. Our analysis shows that the road mostly avoids ancient settlements. Only Jurniyye on the northern branch and two smaller sites (farms?) on the southern branch (Hartal and Ben Efraim 2012b: Site Nos. 116 and 118) stand along the road(s). All other known settlements are located at a distance of at least several hundred metres from the road.

In several places the road is apparently disturbed by later routes (Fig. 11)—north of the Revayyah reservoir and west of the branching, and to the east of ʿUyun Hadid (Hartal and Ben Efraim 2012b: Site No. 118). Both represent medieval or recent roads visible on 19th/20th-century topographical maps (Pažout 2023). At ʿUyun Hadid and on the Lawiye spur (Hartal and Ben Efraim 2012a: Site No. 87), the road is overlaid by animal pens and Bedouin graves.

Fig. 11: The Roman road (running from east to west), probably disturbed by later north–south communication (photo by A. Pazout)

Discussion

Road hierarchy

The construction methods used in the road sections that were studied are consistent with what is known about Roman road construction in the eastern provinces (Bauzou 1985: 145–147; 1998: 109–129; Porat 2007). We may note a narrower width of the Golan road (ca. 5 m) compared to major regional roads (Via Nova around 5.9–6 m; Bauzou 1998: around 6 m; Porat 2007; Bauzou 2003: 296–301); ca. 20 feet as opposed to 17/15.5 feet. This might suggest a hierarchy of roads in the provinces, where the Golan road is clearly secondary in importance. Given the different widths and courses, there are two distinct roads: a wider one (R1), equipped with milestones and watchtowers, that connects the Sea of Galilee with Nawa, and a narrower road (R2), branching away from the former at the Revayyah reservoir. The former is probably the one that appears in the rabbinical sources as the ‘Road of (Beth)-Saida’ (ןדייצד אטרטסא; Leibner 1999: 193–198).

Although no remains of local roads connecting the imperial road (R1) and its southern branch (R2) with local villages were found, it is likely that such roads existed. However, since the imperial road does not connect local settlements, Lapin (2001: 87–94) concluded that economic motivations probably did not play a role in the planning of the road. Nevertheless, we should bear in mind that the Roman Imperial highways were built mainly due to military-strategic considerations and that the connectivity of the rural sites was not their primary objective. The local topography was the main factor in determining the route of the road and was more important than its proximity to villages. The major consideration for the route was the most direct passage between destinations. We cannot exclude the possibility that in the case of the Golan, the fact that the road does not approach any of the Jewish settlements may have been deliberate, perhaps due to security concerns. The Roman (paved) road significantly improved mobility in the winter months. A priori, one would expect the new Roman road to contribute to the motivation of expansion and the possible foundation of new villages in its proximity. As of the late 3rd century, however, the opposite was the case: there was a decrease in settlements in the southern and central Golan, including those near the road such as el-Qusayyibe and Majduliyya.

Road construction and abandonment

The milestones of Marcus Aurelius and Lucius Verus belong to a widespread milestone series of 161/2 CE erected in preparation for the Parthian war (Isaac 2015; Roll 2009). The series might indicate both repairs and new construction; therefore, the road was paved in 161/2 CE or perhaps slightly earlier. The road connects two principal longitudinal Roman Imperial highways: (1) a road from Caesarea Paneas in the north to Scythopolis with (2) a road from Damascus southward, presumably to Adraha-Deraʿ, where it branches east to the legionary camp at Bostra and south to Philadelphia—both cities along the Via Nova (Bauzou 1985: 139–145; 1998: 236–244). At the same time, a road connects these two longitudinal roads: leading from Tyre on the coast through Caesarea Paneas to Damascus, it was also (re-)paved (Roll 2009: 14*). The paving of the road through the central Golan and Caesarea Paneas can be seen as part of the effort to improve connectivity between the Mediterranean ports (Tyre and presumably Akko-Ptolemaïs) and inland Syria in 161/2 CE. The Golan road continued to be maintained at least until the late 3rd/early 4th century CE. The results of our excavations suggest that the watchtowers were erected in the 3rd century and abandoned in the first half of the 4th century CE. It would be tempting to link the erection of the watchtowers to the construction of the road in the second half of the 2nd century, but the archaeological evidence at hand does not permit us to confirm this connection with any degree of certainty. The abandonment of the watchtowers might be linked to an overall rearrangement of the Roman army in the provinces during the reign of Diocletian and later (Wheeler 2007: 252–253).

Watchtowers and military-administrative implications

From the 1st to the 3rd century CE a system of detached service of soldiers (i.e., soldiers serving away from their home units) developed in the provinces. The soldiers were often placed in stationes (stations), in praesidia (fortlets), or simply in watchtowers (Fuhrmann 2012: 201–203). A building inscription from Moesia Inferior, dated to the reign of Antoninus Pius (151/2 CE), refers to construction of praesidia, burgi and phrouri ‘for the protection of the province’ (Gerov 1989: 211). Praesidia are documented from Egypt’s Eastern Desert, identified with large fortified structures; their personnel were sometimes called curatores (Maxfield 2005; Reddé 2018). These structures were generally located on the main roads or in areas with specific security needs (Bishop 1999). A praesidium at En-Gedi is known from the Babatha archive (P. Yadin 1.11). Burgi and phrouri can be equalled to smaller installations, such as watchtowers or blockhouses. Burgi in the sense of watchtowers that were located along roads connecting main cities are mentioned in some Late Antique rabbinical sources (Isaac 1992: 178–184). Some stationes could have been connected to the official post system, cursus publicus (France and Nelis-Clément 2014: 118–120; Kolb 2016). The personnel involved included regionarii—centurions sent to outposts and responsible for policing a given district (regio), often with several soldiers under their command. These were called stationarii, exercising surveillance and basic police functions. They are documented in abundance on ostraca from roads of the Eastern Desert in Egypt (Brélaz 2005: 258–261; Fuhrmann 2012: 207–216; France and Nelis-Clément 2014: 125, 127; Petraccia 2016). Soldiers of various ranks on detached duty would be seen mainly on the important communications and in areas with considerable security risks, predominantly outside city territories (i.e., in non-urbanised districts) and in the frontier fortifications.

It should be stressed that civic (and village) police institutions are not well attested in Judaea, Arabia, or southern Syria, with the evidence limited to Asia Minor and Egypt (Brélaz 2005; Aubert 1995). A distinction may be made between police functions executed by the Roman provincial administration (and the army) and by cities, since they were responsible for maintaining public order in their civic territory. Nonetheless, during the 2nd–3rd centuries CE, soldiers were increasingly deployed to perform police functions in local communities, which had begun to rely to a greater degree upon the governor and his staff (Brélaz 2008: 48; Fuhrmann 2012: 45, 173–177). It seems that there was no sharp divide between legal jurisdiction and policing activities of governors and cities in the Roman period, and this allowed for the flexible operation of both apparatuses (Brélaz 2008).

In the region under study, there are three watchtowers located on the Roman road (R1): el-Qusayyibe East, Maʿoz’s tower and the newly identified tower at Height Spot 500. Here we argue that they represent a system built and operated by the Roman army, rather than by the closest polis of Hippos’ civic authorities, on the grounds of the following: 1) uniform layout and construction methods; 2) equal spacing, using a one-mile interval module; of use in the 3rd century and abandonment in the first half of the 4th century; and 5) the fact that they are located on a newly built road of transregional strategic importance, as described above. The rather short lifespan of the watchtowers is perhaps a strong indicator of their Roman army origin. Presumably, if they had been maintained by the civic authority, they would have been maintained into the 4th century and even later. The road, therefore, can be described as a ‘military’ or limes road (Isaac 1988; 1992: 101–140; Bauzou 1998: 135–140; Baatz 2000: 42–49).

The central and western Golan, just north of the road, were areas of significant Jewish population in the Roman period (Ben David 2005). Jewish villages were also found to the south of the road (Majduliyya and Umm el-Qanatir; Eisenberg and Osband 2019; Osband, Ben David and Arubas 2020), and it is agreed that they belonged to the territory of Hippos (Maʿoz 1986). We may speculate, therefore, that the Jewish population was deemed to be potentially hostile long after the Great Revolt, the Bar-Kokhba revolt⁸ and the establishment of direct Roman control, and that this warranted precautionary measures. To this we may add the lack of any urban centre in the vicinity that could serve as an additional regional security buffer—Hippos is located 15 km away as the crow flies and over five hours’ walking distance (Pažout and Eisenberg 2021: 13).

Unfortunately, there are no 2nd- or 3rd-century historical sources that might shed light on this issue. Nonetheless, the Roman provincial government took precautions to secure lines of communication in this region. The road through Leja in southern Syria might perhaps provide the best parallel for such a security arrangement. In the Early Roman period Trachonitis (i.e., Leja) was notorious for banditry, necessitating intervention by Herod the Great (Josephus, War 1.398–400; Ant. 16.285, 347–348). Strabo, in the early 1st century CE, writes that the banditry ‘happens less frequently’ thanks to Roman soldiers stationed in Syria (Strabo, Geography 16.2.20). We may hypothesise that security issues continued into the 2nd century, when the Romans decided to build a road through Leja, as the shortest route between Damascus and legionary camp in Bostra, and developed a system of watchtowers along the road, with a garrison in Mismiyyeh (Phaena) in the northern tip of Leja (Bauzou 2003: 296; Sartre-Fauriat and Sartre 2014: 33–37). Milestones and other epigraphic evidence date the Leja road and its garrisons between Marcus Aurelius/Lucius Verus and Commodus (161/2–186/7 CE). Similar security arrangements can be identified also on the Caesarea Paneas–Damascus road, also dated by milestones to 161/2 CE, where two watchtowers were identified east of Caesarea Paneas (Roll 2009: 14*; Hartal 2017a: Site No. 81; 2017b: Site No. 69).

Securing lines of communications with watchtowers and direct army involvement in the Golan can mirror the Roman response in Trachonitis. Both regions were recently conquered territories with indigenous populations that were historically hostile to the Roman rule, and both lie on important transregional routes. It seems logical, then, that the Roman army invested in security in these regions.

Ancient landscape

Another issue that needs to be addressed is the relation of the road to the visible field systems. As noted during the survey, many field walls along both the northern and southern branches of the road overlie the road curbs; in other words, the development of field systems along certain segments of the road post-dates the road construction. In many cases, these plots have the regular rectilinear shape of the ‘coaxial’ type, commonly dated to the Late Roman period (Gibson and Lewis 2023: 198–201). These systems, in our view, differ from irregular or circular enclosures and plots that can be associated with 19th/20th-century settlements (ibid.: 219). The settlement density peaked in the Roman period and then again in the Mameluke period (Hartal 2012). There is also no indication of continued use of the Roman road after the Byzantine period, especially after the earthquake of 749 CE. The main road from Damascus was reoriented on a northeast–southwest axis, passing through Jukhadar, Khisfin, Fik and Khan el-ʿAqabeh. In this context we are inclined to date these field systems to the Roman-Byzantine period, but at this point it is impossible to narrow down the dating any further.

Furthermore, there are indications of Roman veteran colonists in Khisfin (south of R2; Gregg and Urman 1996: 74–88; Ben David 2005: 222–223; Hartal 2006: 427–428) and Hippos and its territory (Łajtar 2014: 256–257). In southern Syria, there is tentative evidence for large-scale limitatio—land surveying and land division (Kleb 2014). The two engineered Roman roads described in this article were laid out using Roman surveying techniques and are continuations of roads in southern Syria (Schumacher 1889: 18–20, 37–40; Bauzou 2003); they are therefore probably part of the same land-surveying system. The field systems previously studied at Farj (ca. 7 km north of the Roman road) and Mazraʿat Quneitra in the Golan are thought to have been laid out by land surveyors following Roman models (Gibson and Lewis 2023: 204–205). In this context, the road and associated land division would have been of great local economic importance. We therefore tentatively suggest that the land divisions aligned to Roman roads in the Golan are remains of a Roman limitatio; its remains should be studied further in the context of the rural (veteran?) colonisation in the Roman period in the Southern Levant. Recent research on the Roman settlements in the region of Gadara suggest similar rural colonisation processes in the neighbouring region (Olsvig-Whittaker, Leiverkus and Soennecken 2023).

Spatial analyses

Finally, we must address the contribution of spatial analysis to our research. While the CFMN analysis was designed to provide a movement model for the southern Golan, we decided to use its results as a basis for research on the western terminus of the road. The CFMN algorithm employed is slope-based, excluding very steep slopes and the bottoms of valleys. Therefore, as expected, the modelled route follows the flat top of the Lawiye spur until its western edge, where it starts its gradual descent. This route is corroborated by milestones found on the northwestern terrace of the Lawiye spur and by historical routes in the area recorded in British Mandate 1:20,000 topographical maps (Fig. 12).

Fig. 12: The Roman road system in relation to the historical road system (based upon British Mandate 1:20,000 topographical maps); the Subahiye Ascent is marked in white

The visibility analysis helped us to formulate research questions regarding the intervisibility of watchtowers along the R1 in preparation for the field survey. The survey could not, however, substantiate our premise that the watchtowers were intentionally built for intervisibility (and that elevated positions along the road were selected, for example). The only watchtower securely identified in our new survey is located far from the others and is not intervisible with them; it thus failed to provide any insight into the matter. Short distances (of around 1,500 m) between the watchtowers should accommodate good intervisibility, but as the model showed, some elevations and rolling terrain might obscure visibility even with such short distances. Therefore, it cannot be concluded with any degree of security whether the intervisibility of the watchtowers was intentional or not.

Conclusions

In this article we have presented the results of a multidisciplinary study of a Roman road in the southern and central Golan Heights, covering diverse methodologies, from spatial analysis to surveys and excavations. By combining new and heritage data we have proposed a new synthesis on the Roman military involvement in the region and its historical and administrative context. The epigraphic and archaeological evidence indicate that the road was built in the 2nd century CE, with milestones erected on behalf of Marcus Aurelius and Lucius Verus in 161/2 CE, as part of a larger project of roadbuilding and maintenance, possibly in preparation for the Parthian War (163–167 CE). The road served as an additional east–west connection between the Mediterranean ports and inland Syria.

We found no correlation between the route of the Roman road and the nearby Jewish settlements. This may have been somewhat dictated by the topography or was perhaps a deliberate security-based consideration. The system of watchtowers along the road was 90cee8 developed either in concordance with the 2nd-century road construction or more securely can be dated in the 3rd century CE. In either case, the system was planned and coordinated as a whole, given the uniform design and date of use of the excavated watchtowers. Despite the incomplete nature of the evidence, the newly identified watchtower suggests that the system of watchtowers spaced at regular 1-mile intervals covered the length of the road from the Sea of Galilee to Nawa (ca. 39 km or approximately 25 miles). Its abandonment might be related to large-scale military reorganisation between the reign of Diocletian and Constantine. The direct involvement of the Roman army in policing and surveillance in the interior of the province suggests serious security concerns, felt in a predominantly Jewish area (Gaulanitis) even several generations after provincialisation of the region. The road probably formed an axis of land division (limitatio) in the Golan and southern Syria and played a role in Roman rural colonisation, as suggested through the observed relationship between the road and rectilinear field systems surrounding it. Similar field systems in Judaea, Samaria and Golan were dated to the Late Roman period, but further research is needed to elucidate the date of the field system and its relation to the Roman and Byzantine settlements.

Through the integration and analysis of archaeological data with spatial analyses we have been able to propose a course of the western continuation of the road and its eventual terminus on the shore of the Sea of Galilee. Furthermore, visibility analysis has proven to be a valuable tool for formulating research questions and establishing prospective areas for field survey and validation. Further integration of spatial analytical methods with field research in the study of Roman roads, and especially their context in the landscape, is desirable, as our case study has aimed to demonstrate.

Disclosure statement

The authors report that there are no competing interests to declare.

Additional information

Notes on contributors

Adam Pažout

Adam Pažout: The Zinman Institute of Archaeology, University of Haifa, Israel and Centre for Urban Network Evolutions (UrbNet), Aarhus University, Denmark

Michael Eisenberg

Michael Eisenberg: The Zinman Institute of Archaeology, University of Haifa, Israel; email: [email protected]

Mechael Osband

Mechael Osband: The Zinman Institute of Archaeology, University of Haifa, Israel; email: [email protected]

Notes

1 The administrative boundaries in the Southern Levant in the Roman–Byzantine period have been extensively surveyed (Avi-Yonah 1966; Safrai 1980; Hartal 2006: 353–365). While the northern boundary of the Hippos territorium in the southern Golan is contested, it is generally considered to lie north of Naḥal Samakh (Ben David 2019: 290). Recent research on settlement patterns and connectivity in the region suggests that it lay near the Roman road (Pažout and Eisenberg 2021).

2 For the Roman road system in the Golan, see Schumacher 1888: 61–62; 1920; Urman 1985: 112–113; Maʿoz 1993: 537; Roll 2009. For the previous surveys and excavations along the road under scrutiny, see Hartal 2012; Hartal and Ben Efraim 2012a: Site No. 79; 2012b: Site Nos. 103, 104, 111 and 114; Maʿoz 1982; Zingboym 2008; Tepper and Tepper 2013: 134–135. The milestones catalogued by the Israel Milestone Committee along the road are: IMC 212, 250, 293, 408, 409, 415, 417–419, 423, 464 and 468 (https://milestones.kinneret.ac.il/en/info-about-milestones/north/). For newly published relevant milestones (IMC 212 and 250), see Staab, Pažout and Eisenberg 2020.

3 One Roman mile equals approximately 1,480 m (but compare Bauzou 1998: 133–134).

4 CFMN is based on directionless movement from the source point (Fábrega-Álvarez 2006). It computes the least-cost path for a source point that shows all easiest accessible routes for a given point (focal mobility network). These focal mobility networks for all points are then combined, and their density is computed in a 50 m search radius. Where mobility networks for multiple points converge, there they form natural corridors of movement (Déderix 2016). The CFMN was computed for 51 points spread evenly across the study region, using r.walk module in GRASS GIS. R.walk computes anisotropic cost of movement (direction dependent).

5 Co-directed by Micheal Eisenberg, Mechael Osband and Adam Pažout, Israel Antiquities Authority excavation permit no. G-4/2019.

6 Chaim Ben David, personal communication.

7 Available at GovMap portal, https://apq9h.app.goo.gl/WBdN

8 Although it is doubtful that the revolt and related military operations reached the Galilee and the Golan (Mor 2016: 152–173).