Three thousand years of river channel engineering in the Nile Valley
Scientific editing by Kevin Walsh.
Abstract
Across a 1000-km stretch of the River Nile, from the 1st Cataract in southern Egypt to the 4th Cataract in Sudan, many hundreds of drystone walls are located within active channels, on seasonally inundated floodplains or in now-dry Holocene palaeochannel belts. These walls (or river groynes) functioned as flood and flow control structures and are of a type now commonly in use worldwide. In the Nile Valley, the structures have been subject only to localised investigations, and none have been radiometrically dated. Some were built within living memory to trap nutrient-rich Nile silts for agriculture, a practice already recorded in the early 19th century C.E. However, others situated within ancient palaeochannel belts indicate construction over much longer time frames. In this paper, we map the distribution of these river groynes using remote sensing and drone survey. We then establish their probable functions and a provisional chronology using ethnoarchaeological investigation and the ground survey, excavation and radiometric dating of the structures in northern Sudan, focusing on the Holocene riverine landscape surrounding the pharaonic settlement of Amara West (c. 1300–1000 B.C.E.). Finally, we consider the historical and economic implications of this form of hydraulic engineering in the Nile Valley over the past three millennia.
1 INTRODUCTION
Across a 1000-km stretch of the River Nile, from just above the 1st Cataract in southern Egypt to the 4th Cataract in Sudan (c. 23°50′N–18°50′N), many hundreds of drystone walls built of local stone are located within active channels, on seasonally inundated floodplains or in now-dry Holocene palaeochannel belts. These walls are generally straight and range from several metres to over 200 m in length, rarely exceeding 1 m in height or 2 m in width. Most project partway into the river, perpendicular to flow, while some traverse entire channels.
These walls may be characterised as a type of river groyne—or spur dike/spur wall—structures now commonly used in rivers worldwide (and from the 16th century C.E. in China; Zhaoyin & Cheng, 2019, pp. 36–38) for purposes that include encouraging sedimentation, limiting bank erosion, improving aquaculture and training river channels for navigation or flood defence (e.g., Przedwojski et al., 1995; Savić et al., 2013; Yossef, 2002).
Research on the human modification of channels along the main Nile has primarily focused on the impacts of dam and canal construction in the modern era (e.g., Bunbury et al., 2023; Stanley, 1996; Woodward et al., 2022) or on ancient canalisation and harbour infrastructure (e.g., Boraik et al., 2017; Quirke, 2009). To date, river groynes in the Nile Valley have been subject only to localised examinations (e.g., Tahir & Sadig 2014; Vercoutter, 1966; Wolf & Gabriel, 2008) and, to our knowledge, none have been independently dated. Some were built within the living memory of the region's present-day inhabitants to trap fertile Nile silts for cultivation (Figure 1), a practice already noted by European travellers to Nubia in the early 19th century C.E. (Cailliaud, 1826). However, similar structures situated within ancient palaeochannels indicate construction over much longer time frames, whether for the same or other purposes.
To map the distribution of these groynes and to establish their probable functions and a provisional chronology, this paper utilises wide-scale remote sensing and the ethnoarchaeological investigation, drone survey, ground survey, excavation and radiometric dating of the structures in Sudanese Nubia, especially those located in the Holocene riverine landscape surrounding the New Kingdom pharaonic settlement of Amara West (c. 1300–1000 B.C.E). The results of these investigations inform our discussion of the historical and economic implications of the use of these structures in the Nile Valley from the second millennium B.C.E. to the present.
2 GEOGRAPHICAL DISTRIBUTION
Published records
The northernmost known examples of these structures were recorded just above the 1st Nile Cataract at Debod (Figure 2) and were inscribed alongside others in a map of Lower Nubia produced by the British architects Henry Parke and Joseph John Scoles (1824; Figure 2a). Some river groynes in this area were also described in detail by Johann Ludwig Burckhardt (1819, p. 11). Apart from very cursory references (e.g., Willcocks, 1899, p. 31), we know of no other records of these structures in northern Lower Nubia, with most walls seasonally flooded and/or silted up by the construction and successive raisings of the Aswan Low Dam between 1899 and 1929 (see Trigger, 1965, pp. 37–40). Upstream into the 2nd Cataract and Batn el-Hajar (the rocky, cataract-strewn region between the 2nd Cataract and Dal), groynes were recorded at Mirgissa, Semna and Askut (Vercoutter, 1966), from Sonki to Akasha (Maystre, 1980) and at Kulubnarti (Adams, 2011) (Figure 2e). All now lie beneath Lake Nasser/Nubia (the reservoir of the Aswan High Dam) and are unavailable for further research. Upstream of the reservoir, the structures were recorded from Dal to Amara West (Vila, 1975, 1977b, 1977c), Abri (Burckhardt, 1819, p. 56), Soleb (Schiff Giorgini et al., 2003), parts of the 3rd Cataract (Osman & Edwards, 2012; Tahir & Sadig, 2014), Kerma (Lepsius, 1849–1859, p. 245), Kawa (Macadam, 1955, Pl. CXII d), the Dongola Reach (Welsby, 2001) and the now-flooded 4th Cataract region (Wolf & Gabriel, 2008) (Figure 2e).
Remote sensing survey
Systematic analysis of a variety of high-resolution remotely sensed imagery was undertaken between Aswan and the head of the 4th Cataract (Figure 2). Where recent satellite imagery of the Nile was available, all currently active channels, and all known and hypothesised palaeochannel belts and floodplains, were examined in Google Earth at 1 km 'eye altitude' (c. = 0.75 m/pixel on-screen resolution). Some 500 interpreted river groynes were identified across the survey area (e.g., Figure 2d). Using the same approach, investigation of the Dongola Reach's very extensive palaeochannel network (Macklin et al., 2013; Welsby et al., 2002; Woodward et al., 2001) was attempted but soon abandoned, as extensive agricultural development (Welsby, 2001) has rendered any surviving examples impossible to differentiate from modern irrigation ditches and palm frond sand defences without ground survey.
In areas flooded by the Aswan dam reservoirs, two historical sources of high-resolution aerial imagery were georeferenced and surveyed in the same manner. In aerial photographs taken by the British Geographical Section, General Staff (National Collection of Aerial Photography, n.d.) in May 1934 (Figure 2b), 15 possible groynes were identified between Aniba and Abu Simbel in Lower Nubia. Over 600 were identified from Gemai to Dal in the Batn el-Hajar in 1960s aerial photographs taken as part of the Nubian Salvage Campaign (Figure 2c). Systematic survey was not possible between the Aswan Low Dam and Derr (flooded by the Low Dam reservoir in the 1934 photographs) and from Abu Simbel to Gemai (no suitable imagery available). In total, 1281 interpreted groynes were identified in satellite/aerial imagery or in published records (Figure 2).
3 GROUND SURVEY, EXCAVATION AND ETHNOARCHAEOLOGICAL INVESTIGATIONS
Introduction
To reconstruct the chronology and function of these structures, a programme of ground- and drone-based survey was instigated in northern Sudanese Nubia in January 2019, alongside excavation, radiometric dating and ethnoarchaeological investigations. Fieldwork focused on walls located in an archaeological concession surrounding the New Kingdom pharaonic town of Amara West. This concession was granted to the British Museum's Amara West Research Project by the National Corporation for Antiquities and Museums, Sudan (NCAM), and spans some 9.4 km of the left bank of the Nile at the upstream end of the Abri–Kosha reach, a 20 km stretch of the river that flows from west to east (Figures 2e and 3). The concession shares its boundaries with a present-day omodiya (administrative district), also called Amara West.
The pharaonic and post-pharaonic remains of Amara West and its environs have been under investigation by the multidisciplinary Amara West Research Project since 2008 (Spencer et al., 2014). The Late Holocene palaeochannel system in this reach has also recently been studied in detail, creating a chronological and geoarchaeological framework (Woodward et al., 2017) and providing valuable context for an investigation of the walls in this area.
Walls were also briefly surveyed in the nearby villages of Dal, Quweyka and Wawa, and with the collaboration of the Section Française de la Direction des Antiquités du Soudan (SFDAS) in the concessions of Sai Island, Soleb and Sedeinga (Figure 2e).
Ethnoarchaeological perspectives
While ground surveying walls in the villages of Quweyka (see below) and Suwarda, several local farmers were informally interviewed in their fields. They identified these structures as al-sudud (dams),1 constructed during their fathers' or grandfathers' generations. According to the farmers, the structures were built for two purposes: to capture Nile silt during the annual inundation and to prevent the erosion of existing channel margin alluvial silt deposits.
After the flood recedes, fertile and saturated silty soil (saluka) provides agricultural land that requires only limited, if any, mechanical or manual irrigation. Saluka land is divided into two categories by modern farmers in Nubia: jarf, the lower, sloping riverbank areas, and gurer, the higher, flat terraces that were also seasonally inundated but, due to lower floods since the completion of the Merowe Dam in 2009, must now be mechanically irrigated (Ryan et al., 2022).
These river groynes thus improve agricultural potential by increasing the amount of cultivable land and by protecting existing alluvial soils. Even though they have not been built in recent decades in these areas, the jarf land formed behind some walls remains in use (Figure 1).
One respondent told us that his father hired workmen to build a wall on a small island called Sermutta, adjacent to Quweyka (Figures 2e and 11). This prompted sedimentation that caused the island to grow, after which stone from the original structure was moved to create more walls, enlarging the island further. The ends of three walls remain visible along its eastern bank, protruding from silty flood deposits. Comparison with declassified Corona spy satellite imagery from February 1969 indicates that the island has since grown from 4.2 ha to its current 5.7 ha, with most additional silt deposited along its eastern and downstream banks. Due to its steep banks, Sermutta is unsuited to jarf cultivation. Since the mid-1960s, the island's interior, which only floods during exceptional inundations (last in 1991), has been planted with date palms irrigated first by a sagia (wooden, cattle-drawn waterwheels) and now by diesel pumps.
Within living memory, groynes have also been built to capture and retain silt elsewhere in Sudanese Nubia (Tahir & Sadig 2014, p. 49). They were evidently built at Kulubnarti during the 20th century C.E. (Adams, 2011, p. 65) and as early as the 1820s in other parts of the Batn el-Hajar (Cailliaud, 1826, p. 351). The 20th century C.E. construction of the structures for these purposes is also attested across the Batn el-Hajar more generally (Vercoutter, 1966, p. 153).
Ground survey method
Ground-surveyed walls in the Amara West Research Project concession were assigned a unique archaeological context number, with those surveyed elsewhere labelled with the location and a sequential number suffix. All were photographed, and maximum visible heights and representative widths were measured (Figure 6). Wall lengths were measured from either drone orthophotos (Amara West, Soleb, Sedeinga and Wawa) or high-resolution satellite imagery (Dal, Sai and Quweyka). The rock type, size and shape of the building stone were also recorded.
A simple typology of construction method was created and assigned to surveyed walls for which this was identifiable. Two types comprise outer walls with vertical faces of drystone masonry infilled by smaller rocks and Nile silt. In the first, stones are bedded flat, with any elongate stones dominantly positioned parallel to the wall's length (Figure 4a). In the second, stones are bedded upright, with elongate stones laid perpendicular to the wall (Figure 4b). A third type encompassing (mostly very wide) 'homogeneous' walls comprises rocks of similar size throughout the structure (Figure 4c), while a fourth type comprises a single row of stones only (Figure 4d).
Surveyed structures were also classified if the construction method could not be identified. The first indeterminate type comprises linear mounds of loose stone with no visible masonry (Figure 4e). While some structures may have been constructed in this manner, one excavated example at Amara West (1300; Figure 9) revealed a buried wall with outer faces and fill. As such, all such mounds are considered to be of indeterminate construction method, pending excavation. Finally, c. 20% of surveyed structures appeared only as linear scatters of loose stone on the modern land surface (Figure 4f). These are presumably the deflated remains of former walls.
Archaeological investigations: Amara West
3.4.1 Historical and environmental contexts
In addition to its New Kingdom town, the Amara West Research Project concession contains rich, multiphase archaeological remains in a wider desert landscape consisting of plateaus of Precambrian rock cut by wadis and well-preserved palaeochannels of the main Nile, covered in places by thick deposits of aeolian sand (Woodward et al., 2017). These landscapes were the subject of ground survey and limited excavations led by André Vila in 1972–1973 (1977c), investigations subsequently enhanced by targeted excavations of the Amara West Research Project (Stevens & Garnett, 2017) and the Research Unit for Later Prehistory, Sai Island Archaeological Mission, University of Charles de Gaulle-Lille 3 (Florenzano et al., 2016; Garcea et al., 2016).
Although the site dates ascribed by the 1970s survey require refinement through further fieldwork and analysis (see Budka et al., 2019), the principal patterns of human occupation in the area can be outlined. The first evidence dates to 8700–8500 cal. B.C.E. and attests to the foraging Arkinian cultural horizon living within a savannah environment, followed by Khartoum Variant sites (c. 7550–4850 B.C.E.). For both periods, hearths and lithic scatters can be identified, with pottery also present within the Khartoum Variant sites. The onset of aridification from the mid-sixth millennium B.C.E. saw a shift in human activity towards river channels, with a spread of Abkan culture sites marking the transition from the Mesolithic to Neolithic and evidence of domestic cattle and, later, caprine herding (c. 5500–3500 B.C.E.), alongside the introduction of cereal agriculture around 4000 B.C.E. The ensuing Pre-Kerma culture and Kerma culture (c. 3500–2500 and 2500–1500 B.C.E., respectively; Schrader & Smith, 2021) are represented by cemeteries but also a number of sizeable settlements, some of which continued in use into the New Kingdom (c. 1550–1070 B.C.E.), the era of pharaonic Egyptian colonisation and rule over Upper Nubia. Dating to the earliest period of Egyptian occupation (15th century B.C.E.), these may have been associated with the control of desert routes and the extraction of resources, presumably overseen from the major pharaonic centre upstream at Sai island (Budka et al., 2020).
Around 1300 B.C.E., a new pharaonic centre was established at Amara West (Spencer et al., 2017), comprising a walled town with cult temple, storage and workshop facilities and housing areas, but also the residence of the Deputy of Kush, the foremost pharaonic official in occupied Nubia. Inscriptions, seal-impressions, storage and production areas attest to the collection, transformation and onward supply of natural resources, particularly gold. The inhabitants of Amara West were buried in two nearby cemeteries (Binder, 2017) (Figure 5).
It has long been recognised that the walled town of Amara West was founded on an island (Spencer, 1997, p. 1), bordered immediately to the north by the c. 140 m wide 'Amara West' palaeochannel (Spencer et al., 2012; Woodward et al., 2017) (Figure 5). This forms part of a 'southern' palaeochannel belt that runs parallel to the modern Nile, upstream and downstream of the pharaonic town (Figure 3). To the north, a series of anabranching palaeochannels runs across the desert (Figures 3 and 5). This 'northern' channel belt would have diverged from the main Nile just upstream of the New Kingdom town (Woodward et al., 2017), flowing northeast through a rocky cataract and thereafter through the omodiyat of Attab West and Ginis West.2 When active, the northern palaeochannel belt was up to 1 km wide and circumscribed a large island some 6.5 km long and 1.8 km wide to the south. Walls are preserved in both palaeochannel belts, with those situated closest to the pharaonic town concentrated in three discrete clusters (Figure 3). These walls were first surveyed by Vila (1977b, 1977c), who assigned them the site designations used here. Vila ascribed the structures to the 'Nubian Moyen–Kerma' (c. 2500–1500 B.C.E.) based on adjacent settlements dated to these periods. This interpretation agrees with that of Anthony John Arkell, who surveyed these walls in 1939 and associated them with 'C-group cultivators' (Vila, 1977c, p. 24).
Optically stimulated luminescence (OSL) dating of the northern palaeochannel belt and Amara West palaeochannel indicates that both had stopped flowing perennially/seasonally by 1270 B.C.E. (date range 1485–1055), that is, within a generation of the New Kingdom town's foundation (Woodward et al., 2017). Thereafter, these channels experienced only very rare, low-energy inundations in years of exceptional Nile flooding, with flow ceasing permanently some time before 205 B.C.E. (date range 385 B.C.E.–25 C.E.) (Spencer et al., 2012, fig. 3; Woodward et al., 2017).
When flowing, these anabranching channel systems would have provided effective barriers to the sand transported by prevailing northern winds since at least the later New Kingdom (Munro et al., 2012; Woodward et al., 2017). The drying out of these channels would have had a profound effect on lifeways at the pharaonic town of Amara West, most significantly through the loss of agricultural land and the multiple impacts of increased sand ingress (Woodward et al., 2017). Ultimately, the viability of settlement here became too challenging, and the town was abandoned around 1000 B.C.E. if not somewhat earlier.
This inhospitable environment may explain the limited evidence of first millennium B.C.E. (Napatan) (see also Budka et al., 2019, pp. 20–21) and Medieval activity across the left bank of the reach, including 8th century B.C.E. burials within Amara West's cemeteries. From the first millennium C.E. onwards, the opposite bank has been the focus of settlement, agriculture and burial (see Vila, 1977a, 1977b, 1977c). Twentieth and twenty-first century activity on the left bank was until recently largely restricted to small-scale, diesel pump-irrigated agricultural schemes (see below), but mechanised gold prospecting and mining over the past decade has seriously impacted this landscape.
3.4.2 Drone survey and flood modelling
Drone photography was captured over a 9.1 km2 area of the Amara West Research Project concession, from which a 0.05 m/pixel orthophoto of its palaeochannel systems was produced using Agisoft Metashape. A 0.25 m/pixel Digital Elevation Model (DEM) of a 4.1 km2 area covering the Amara West palaeochannel and parts of the northern and southern palaeochannel belts (Figure 5) was also produced.
Flow simulations were undertaken using this DEM to contextualise the walls in relation to ancient inundations. This was achieved using two-dimensional (2D) simulations in HEC-RAS 6.1, hydraulic modelling software published by the US Army Corps of Engineers. Steady-state flows at increasing, arbitrary discharges were modelled over this terrain up to 900 m3/s, when flood levels reached the highest wall in site 2-R-14 (1397) (Figure 5). This model corresponds well with archaeological evidence from Amara West for maximum New Kingdom inundations, with all excavated buildings in the pharaonic town located above this level. It is important to note that these models utilise current topography, and their accuracy is therefore impacted by the infilling of former channels by fluvial or aeolian sediment as well as the deflation of riverine deposits.
A similar simulation was undertaken over a 105 km stretch of the Nile Valley,3 using the 30 m resolution ALOS World 3D DEM published by the Japan Aerospace Exploration Agency (2015), augmented by elevation data for the Nile surface extracted from 30 m resolution Shuttle Radar Topography Mission (SRTM) data.4 The latter reflect Nile levels at the time of acquisition (February 2000) but not riverbed bathymetry, meaning that models of real-world flows will yield inaccurately large floods. To approximate maximum floods during periods of wall construction, steady-state flows were modelled at increasing arbitrary increments to 6000 m3/s, when inundation levels matched those outlined above in the same locations. Whilst spatially coarse and subject to the same topographic limitations described above, these modelled flood levels agree well with the placement of walls in known and hypothesised palaeochannel belts and floodplains, with most submerged at peak inundation (Figures 3, 10C, 12, and 15a).
3.4.3 Ground survey
The 259 walls identified in high-resolution satellite imagery within the Amara West Research Project concession were ground-surveyed. Of these, 19 were false positives (e.g., bedrock outcrops, modern ditches or irrigation pipes). One hundred and eighty additional walls were identified during ground and drone survey, bringing the total number of walls located in the concession to 420. This ground-truthing demonstrates the accuracy of wall identification in palaeochannels using satellite imagery (with c. 92.7% being actual structures), but also suggests that a significant number may not be visible in such imagery.
Most walls at Amara West are built of the dark, Precambrian greenschists that dominate the local landscape (Vail et al., 1973). Some structures contain 10%–100% quartz; these are invariably located near outcropping quartz seams, from which this rock was presumably sourced. Building stone ranges in size from cobble (64–256 mm) to boulder (>256 mm), and in form from sub-rounded to angular, and equant to elongate depending on localised bedrock variations. Well-rounded riverine quartz cobbles are commonly integrated into the fills of walls in site 2-S-44. These occur naturally here as scatters across the palaeochannel surface.
The walls at Amara West range from 0.6 to 1.65 m in width (mean = 1.0 m) (Figure 6), with the vast majority of typologically classifiable examples comprising outer drystone faces with an internal fill (Figures 4a,b and 7). Of these, faces of boulders placed flat and parallel to the wall were near universal, except for site 2-S-36, where 25% were built from boulders placed upright and perpendicular to the wall (Figure 7). Narrow, linear stone scatters were common in the north-eastern part of 2-S-44, on the gently sloping mainland bank of the northern palaeochannel belt (Figures 4f, 7, and 10d).
Regardless of construction method, most walls slope down palaeochannel banks and channel margins. The majority are either freestanding or abut rocky outcrops at their higher end only. In site 2-R-14, however, some extend horizontally between bedrock outcrops.
3.4.4 Excavation and dating
To assess the purpose of these structures and retrieve material for radiometric dating, four walls were excavated in different areas of the Amara West Research Project concession. From site 2-R-14, two walls (1343 and 1397) at varying levels within the northern palaeochannel belt were selected and one (1300) within the Amara West palaeochannel, close to the New Kingdom town. One wall (1452) in site 2-S-44 was excavated. Time constraints ruled out excavating walls in site 2-S-36.
To expose standing sections of deposits pre- and post-dating their construction, 1m-wide slot trenches were excavated across the walls, down to their foundations and into underlying deposits. Windblown sand was cleared from both sides but, expeditiously, only the upstream faces of these walls were excavated to full depth. Sand-rich deposits were sampled for multigrain, single-aliquot OSL dating.5
The theory underpinning OSL dating may be summarised as follows (after Feathers, 2008): grains of minerals such as quartz and feldspar act as natural dosimeters for background levels of ionising radiation by trapping electrons within their crystal structure. This signal builds up over time and is zeroed (or bleached) by sufficient exposure to sunlight. By sampling buried sediments in sealed tubes, this trapped charge can be measured in the laboratory by exposing grains to light and recording the intensity of induced luminescence. This value is then compared to dose rates modelled for a given burial environment, indicating the date when sediments were last exposed to sunlight (i.e., the moment of burial). When deposits underlying or abutting a structure can be dated, they provide an earliest or latest possible date for its construction.
Other datable materials (e.g., organic materials or diagnostic artifacts) within abutting deposits or inside the structures themselves may provide an earliest possible date for their construction. As such, the top c. 0.1 m of internal wall fills was also excavated wherever possible to retrieve sealed datable material without disturbing the structure.
Wall 1343
Wall 1343 is 46.2 m long and descends c. 4 m down the sloping bank of a former island within the northern palaeochannel belt at site 2-R-14 (Figures 5 and 8c). This palaeoisland is topped by site 2-R-19, a small settlement with handmade ceramics broadly datable to the Middle Kerma (c. 2050–1750 B.C.E.) and some contemporaneous, imported Egyptian wheel-thrown ceramics (Middle Kingdom, c. 2055–1650 B.C.E.) (Garnett, 2016; Stevens, 2015). Site 2-R-19a, located on an adjacent rise (Figure 8c), has a similar ceramic assemblage (Garnett, 2016). The structure comprises two vertical faces of sub-angular, elongate and equant schist boulders, infilled with consolidated silt and angular cobbles. Where excavated (near its lower terminus), the wall is abutted by 0.75 m of deposits (Figure 8a,b) comprising at least seven, 20–200 mm thick bands of hard, grey, silty units capping fluvially reworked, coarse yellow aeolian sands with fine silty banding. Each sand/silt couplet likely reflects a discrete Nile flood (Macklin et al., 2013; Spencer et al., 2012; Woodward et al., 2017, 2001).
An OSL sample taken from a sandy unit beneath the wall (PS1007) indicates construction after 1205 B.C.E. (age range 1379–1031 B.C.E.). Two OSL samples taken from discrete sandy lenses abutting the wall yielded dates of 700 B.C.E. (PS1017; age range 855–545 B.C.E.) and 440 B.C.E. (PS1014; age range 575–305 B.C.E.). The stratigraphically more recent of these deposits provided an earlier date, but the two date ranges overlap. They indicate that the wall was constructed before the mid-first millennium B.C.E.
Wall 1300
Before excavation, wall 1300 manifested as a 17 m long, 0.6 m high linear mound of silt and schist cobbles and gravels (4–64 mm), situated 200 m west of the walled town (Figures 5 and 16). Excavation revealed a wall comprising two faces of elongate schist boulders densely infilled by angular cobbles and silt (Figure 9). The structure overlaid, and was abutted by, stacked but undifferentiated Nile silt units. Excavation within the wall's fill revealed a c. 20 mm diameter stick of wood charcoal. This presumably derives from nearby occupation deposits, perhaps dumped onto adjacent jarf land and then built into the wall's fill. AMS 14C analysis carried out by Beta Analytic Inc. yielded a date of 1688–1622 cal. B.C.E. at 1σ (c. 68.2%) probability (Bronk Ramsey, 2009; Reimer et al., 2013) (Figure 9). The construction of the wall must post-date the calibrated age of this charcoal.
Walls 1452/1507
Wall 1452 is 24.2 m long and situated in site 2-S-44, on the southern bank of a low palaeoisland in the northern palaeochannel belt (Figures 3 and 10c). It comprises outer faces of angular schist and rounded quartz cobbles with a core of smaller rocks and silt, and is preserved to a maximum of 0.2 m in height. Excavation showed that this wall was constructed on an older wall of angular schist, 1507 (Figure 10a). This lay on a unit of fluvially reworked aeolian sand with fine, silty banding (sampled for OSL dating as PS1000) and was abutted by two silt units separated by a sandy lens (PS1010). Sparse intact plant roots and numerous root voids were observed throughout this sequence.
OSL analysis of the sandy lens (PS1010) yielded a date of 1000 B.C.E. (age range 1185–815 B.C.E.), providing a limit for the construction of the older wall and the earliest possible date for the more recent structure. However, analysis of the stratigraphically earlier sand unit (PS1000) yielded a younger date of 440 B.C.E. (age range 610–270 B.C.E.), indicating that wall 1507 was built after this. Further single-grain OSL analysis is required to investigate whether this discrepancy may relate to the partial bleaching of grains in PS1010 (thereby providing an earlier date) or to the introduction of grains bleached at other times into one or both sampled deposits by root action. As the northern palaeochannel belt was activated only very rarely during the first millennium B.C.E. (Woodward et al., 2017, p. 238), the later date indicated by PS1000 is plausible but judged less likely to be accurate.
Wall 1397
Wall 1397 is 38 m long, 1.4 m wide and is the highest elevation wall in site 2-R-14, at 194.5 masl (Figure 5). It shows outer faces of dominantly elongate, angular schist boulders bedded flat and parallel to the wall with a fill of angular cobbles and silt. Where excavated, the wall is 0.74 m high. It sits on a unit of grey silty sand and is abutted by 0.53 m of stacked but undifferentiated silty deposits containing some animal bone and several undiagnostic potsherds of handmade ceramic vessels (V. Gasperini, personal communication, January 2019). A bone sample was submitted for AMS 14C dating but failed to yield a separable collagen fraction.
Ground survey elsewhere
3.5.1 Sai Island, Quweyka and Sermutta
Eleven walls on the southern tip of Sai Island and one on the adjacent mainland at Quweyka were identified in satellite imagery and ground-surveyed, alongside three others reported to us on the small, nearby island of Sermutta by the aforementioned respondent (Figure 11). All comprise outer faces of dark, metamorphic rocks bedded upright and perpendicular to the wall, with a fill of smaller rocks and silt (e.g., Figures 1 and 4b). In addition to the walls built on Sermutta by his father (see above), this farmer believed that similar structures in Sai Island and Quweyka were built during the 20th century C.E. Those on Sai Island are situated on a broad expanse of silty sand formed against the island's upstream end. These walls are slightly wider than those observed at Amara West (1–1.6 m; mean = 1.3 m).
When visited in January 2019, jarf land abutting the upstream face of the Quweyka wall was planted with unirrigated cowpea (Vigna unguiculata) (Figure 1). A broad expanse of wild grass growing here was also used to pasture sheep and goats. Land surrounding the walls at Sai Island was not cultivated, with agriculture confined to higher, mechanically irrigated gurer land nearby. However, the walls here were abutted by alluvial silt deposits where grasses were also used for pasturing.
Situated amongst the aforementioned walls on Sai Island and located by the edge of the sandbar in January 2019 are a series of low, c. 0.2–0.5 m wide, unfaced mounds of cobbles that extend between bedrock outcrops to form shallow, water-filled basins either contiguous with the Nile or wholly enclosed (Figure 11a).
Further north, four walls were surveyed on the island's eastern banks. All were markedly different from those to the south. Two comprised outer faces of equant quartz boulders with a fill of quartz gravels and cobbles. The others were massive, homogeneous structures built of quartz boulders, most impressively a c. 5.5 m wide mass of c. 0.3–0.4 m boulders within the main Nile channel (Figure 11b). A c. 17 m stretch of this wall was visible in January 2019, but satellite imagery taken at lower water levels shows that it connects to Sai Island and is at least 42 m long.
3.5.2 Dal
On channel margins and in low-lying areas around Dal village, on the west bank of the Nile, 36 walls were identified in aerial photographs taken in the 1960s as part of the Nubian Salvage Campaign (Figure 12). These are the northernmost walls visible in this series of photographs that remain unflooded by the Aswan High Dam reservoir, although those formerly situated on active channel margins are now silted up. Upon ground inspection in January 2019, the 34 accessible examples proved to be walls, suggesting that identifications made elsewhere in the Batn el-Hajar using these photographs are likely to be accurate.
These walls are constructed of unworked local granite boulders and cobbles using a variety of building methods (Figure 7). Excluding narrow single row structures, walls measured 0.7–3.4 m in width (mean = 1.35 m).
All of the walls are situated in areas probably subjected to past inundations (Figure 12), with most also visibly abutted by silty flood units. It is noteworthy that at least one wall rises to the level of c. mid-20th century village houses (Figure 12a), suggesting that it was built during a period of higher flood levels. Excavation and radiometric dating of walls here is a priority for further research, to tie their construction and use into the long-term settlement history of Dal as documented by Vila (1975).
3.5.3 Soleb
Soleb, on the west bank of the Nile, contains a large and well-preserved temple built predominantly in the reign of Amenhotep III (c. 1390–1352 B.C.E.), alongside extensive cemeteries and a mostly unexcavated New Kingdom walled town (Schiff Giorgini et al., 1965, 1972). The temple is positioned downstream of a 1.3 km long cataract of outcropping quartz beds. These are associated with rapids, sandbars and small vegetated islands, as well as a series of large, linear structures visible in aerial imagery but hitherto only briefly described by the site's excavators.
Soleb was selected for ground and drone survey to accurately map and record these structures and to better understand their function. This exercise was constrained by unusually high Nile levels when visited in January 2019. The location of partly exposed or underwater features can nonetheless be inferred by subsurface 'shadows' and lines of rapids visible in drone photographs or other aerial imagery taken at lower flow levels (Figure 13), especially February 1969 Corona imagery.
Today, Soleb temple is located some 500 m west of the Nile. Michela Schiff Giorgini and colleagues reconstructed a sequence of monumental works that provided boat access to temple docks, including canals, basins and an embankment leading to a riverfront wharf (Schiff Giorgini, 1959, 1962). Another apparent jetty of sandstone ashlars ('le massif sur le Nil'; Schiff Giorgini, 1965, p. 4) was situated some 400 m north of the temple's central axis. This structure, which was still visible in the mid-20th century (Schiff Giorgini et al., 2003, figs. 6a and 225a,b), now lies under cultivated alluvium. It is interpreted as part of the northern entrance to an 8 ha enclosed harbour ('digue'; Schiff Giorgini et al., 2003, fig. 224), of which it was evidently the only exposed component.
To the north and south of the temple, several other substantial stone structures were noted within and on the banks of the Nile. Captioned photographs of one example show a meandering wall of boulders extending into the river (Schiff Giorgini et al., 2003, fig. 226a,b). Schiff Giorgini and colleagues produced a tentative reconstruction, in which two sets of parallel stone 'barrages' upstream and downstream of the temple extend from the west bank to a central island, cutting off river access to the digue and creating a diversion channel east of the present-day island (Schiff Giorgini et al., 2003, fig. 6a).
Our survey of these structures reveals a rather more complex picture. Within the cataract, 1 km upstream of the temple, the top of a 4 m wide, homogeneous wall of equant quartz boulders was located on the bank (Figure 13a). The wall extends into the channel, where it is clearly visible as a straight, 110 m long subsurface feature with rapids and occasional protruding boulders. Two hundred metres downstream of this structure, a parallel subsurface feature extends c. 195 m from the bank to a quartz outcrop, where it emerges as a 3.8 m wide wall of identical construction. This bends and continues via several further outcrops to a c. 50 m wide gap of fast-flowing water, bordered to the north by a bedrock outcrop or wall visible in the Corona imagery and now marked by rapids. This connects to an area of quartz outcrops and rocky bars, on which traces of two further walls were recorded. Any further walls or outcrops to the east are covered by a c. 22 ha vegetated island (formed over a sandbar visible in the 1969 Corona imagery). East of this island is a shallow, c. 250–400 m wide backwater channel with no visible bedrock outcrops.
Downstream of the temple, a c. 1.5 m wide stretch of the barrage recorded by Schiff Giorgini was identified in January 2019 (Figure 13). No other walls were identifiable in this area on the ground or in available aerial imagery.
3.5.4 Sedeinga
Sedeinga is located 14 km north of Soleb and contains a temple dedicated to Tiye, built around 1360 B.C.E., and extensive, multiphase cemeteries (Rilly, 2018; Rilly & Francigny, 2010; Schiff Giorgini, 1965, 1966). One wall identified from satellite imagery here was ground-surveyed (Figure 14). This structure is located 250 m east of the temple, adjacent to and partly built on a massive spur of dark, metamorphic bedrock projecting into the main Nile. The clearest segment is 2.9 m wide and comprises boulders of a similar size (c. 0.4–0.8 m) throughout. This spans 8.8 m of a seasonally flooded channel between the outcrop and a cultivated terrace of Nile silt, which also covers its western terminus. To the east, traces of this wall (in the form of scattered, similarly sized boulders) were located on the same orientation atop the bedrock spur. In drone imagery, this bedrock spur and/or wall can be traced a further 100 m or so into the main Nile channel (Figure 15).
3.5.5 Wawa
Eight hundred metres north of the village of Wawa, 10 linear features were identified in satellite imagery on both sides of a shallow depression. These features were ground- and drone-surveyed. They comprise straight, c. 25–60 m long mounds of rounded quartz and sub-rounded basement complex gravels and cobbles with no visible facing, rising to c. 0.4 m above the current ground surface and covered by coarse aeolian sand (e.g., Figure 4f). The features are partly surrounded by a cemetery of some 200 tumuli of 2–15 m diameter (Figure 15). According to our flood modelling, this area would have been inundated under high river discharges (Figure 15a).
Based on tumulus form and surface finds, Lawrence Kirwan dated this cemetery to the post-Meroitic period (1935; cf. Edwards, 2004, pp. 200–207). Tumuli built in this area strongly suggest that any perennial flow or seasonal flooding had permanently ceased before they were built, likely during the first half of the first millennium C.E. Presuming that they served a hydraulic function, the walls were therefore built before this.
4 DISCUSSION
River groynes, irrigation and agriculture
Most structures surveyed would have projected into seasonal floodwaters along the same slope as the bank, perpendicular to flow. When partly submerged, such groynes deflect water and increase adjacent flow velocity, whilst simultaneously slowing or reversing flow along the bank (Copeland, 1983). This has the dual effect of scouring the riverbed at the structures' tips and encouraging silt deposition along the bank. Groynes oriented perpendicular to the bank tend to cause sedimentation on both sides of the structure (Przedwojski et al., 1995). By deflecting flow, these structures also protect pre-existing riverbanks from erosion.
All excavated walls at Amara West showed substantial deposits of fluvially deposited silt against their upstream faces. Where not masked by aeolian sand, all walls surveyed at Amara West, Dal, Sai Island and Quweyka also showed silt banked up against them. This suggests that the structures functioned as effective sediment traps. Like modern examples in Sudanese Nubia, these river groynes were most likely built to increase the amount of arable saluka land available for agriculture following the recession of the annual flood (décrue irrigation). Improved navigation can be discounted as a motivation for construction in all but the few cases outlined below, because most walls would only have been inundated during the flood, when flow was at its deepest.
Much of the Nile Valley between the 1st and 3rd Cataracts is characterised by a narrow floodplain with limited land available for agriculture. Sagia waterwheels, widely used in Nubia from the 3rd century C.E. (Fuller, 2014), provided an efficient means of irrigating higher saluka land (gurer) and extending agriculture into non-flooded barju soils further inland. Before this, flood-irrigated saluka soils would have comprised most available agricultural land, supplemented by watering by hand or shaduf, a mechanical lifting device used in the Nile Valley since pharaonic times (Butzer, 1976). Any method of increasing or preserving saluka land, and extending the period of post-flood cultivation, would have been of great importance to the agrarian societies that lived here. Even after the sagia was introduced, saluka land retained its importance due to the relative ease and cost-efficiency of its cultivation. Sagias required wood to build and cattle to operate, and were therefore resource-intensive.
Wide variations in Nile flood levels would have presented challenges in both designing groyne systems and growing crops on the resultant saturated soils. Flood modelling at Amara West demonstrates that the walls in site 2-R-14 were built across a wide range of elevations (Figures 5 and 8c). When floodwaters lapped against wall 1397, for example, the lower end of 1343 would have been 4.4 m underwater. Across their range, walls were also built down sloping banks. By placing the structures in these settings, their builders ensured that they would continue to trap silt as the flood slowly receded and also function during years of lower inundation.
Recent agricultural practices in Nubia provide a rich body of data suitable for informing past land use in the northern Sudanese Nile Valley (Ryan et al., 2022). Meanwhile, palaeobotanical studies have revealed some of the crop types grown in these landscapes, including during the second millennium B.C.E. (e.g., Budka et al., 2020; Cartwright & Ryan, 2017; Ryan & Spencer, 2013, pp. 336–344). Hulled barley, for example, a staple in pharaonic Nubia, was grown as a key winter cereal until the mid-20th century C.E. on recently flooded jarf and gurer land (including seasonal channel beds) without additional irrigation or with irrigation sometimes required at the end of the c. 80–90-day growing cycle (Ryan, 2016). Emmer wheat, another pharaonic staple, could have been grown in a similar way.
After initial cultivation in saluka areas after the flood, some crops could have been grown without irrigation before the next inundation, due to residual moisture and the proximity of the water table. Today, some low, seasonally flooded jarf land remains cultivated between inundations (Ryan et al., 2022). As they were grown by Amara West's New Kingdom inhabitants (Cartwright & Ryan, 2017, p. 12), watermelon and hulled barley are of particular relevance here. Modern farmers in Nubia noted that watermelon could be planted on jarf land both before and after the flood. Furthermore, whilst barley was formerly planted on jarf land in September, it could also be sown a second time in December/January. The later sowing of barley was used as a fodder, so it did not matter that the grains only partially grew due to the heat and aridity. However, most of the crops sown in these areas after January are arid-tolerant African crops (e.g., cowpea) introduced to this region in post-pharaonic times.
Evidence for post-flood irrigation has been recorded at site 2-T-67 in Attab West, c. 7 km east of Amara West's walled town (Figure 3). Here, Vila recorded three adjacent pits (one stone-lined) connected via subterranean openings to long channels (1977b, pp. 93–96). They are dated, on the basis of a single ceramic bowl, to the New Kingdom. These features were evidently located on the northern bank of a palaeochannel (Vila, 1977b, 93–96), close to a cluster of walls (Figure 3). They may be interpreted as wellheads for shadufs, allowing the irrigation of surrounding saluka soils after the flood. Similar wells have also been noted close to walls on the banks of a large palaeochannel in the 3rd Cataract (Tahir & Sadig, 2014, p. 48). Shadufs continued to be used in Sudanese Nubia until very recently (Ryan et al., 2022, p. 255).
These features prompt consideration of a network of unexcavated, linear depressions located immediately west of the pharaonic town at Amara West, clearly visible in a magnetometry survey and in drone orthophotos (Figure 16). These features are suggestive of irrigation canals. One connects to a pit-like feature with traces of stone lining (Figure 16a), possibly a well from which water was raised before being distributed across adjacent arable land.
Groyne distribution
The vast majority of interpreted groynes have been recorded in narrower, rockier areas of the Nile Valley or in now-dry palaeochannel belts and floodplains. It is important to note likely biases in identification. These stretches of the Nile have changed less over time, in contrast to reaches where the river is highly dynamic within wider floodplains (e.g., Bunbury, 2019), thus potentially burying or destroying groynes. As already noted, it is difficult to identify the structures in satellite and aerial imagery where intensive modern agriculture and settlement is taking place, and they are also more vulnerable to destruction in these areas. As such, groynes were probably more widely distributed across the study area than indicated by our remote sensing survey (Figure 2).
Such issues may explain the dearth of walls identified along the c. 350 km stretch of the Nile between the 3rd and 4th Cataracts. The floodplain here is generally much wider than that to the north and is intensively cultivated, as are numerous adjacent palaeochannel belts, particularly in the eastern part of the Northern Dongola Reach. A number of 'wadi walls' in the Alfreda Nile palaeochannel belt were identified during ground survey (Welsby, 2001, fig. 3.26). These indicate that the technology was used here before the last flows in this channel belt around 280 C.E., but most likely before seasonal flows ended around 1290 B.C.E. (Macklin et al., 2013, p. 697). It is also possible that the extensive, anabranching channel networks of the Dongola Reach provided sufficient saluka land without the need for such walls (Macklin et al., 2013, p. 698). Even if beneficial, a lack of building stone across much of this broad reach may have ultimately precluded construction. The reality (or otherwise) of this apparent gap in distribution is currently impossible to test, but may become clearer with future ground survey and widening access to archives of historical aerial imagery.
To our knowledge, the only potentially comparable structures located north of the 1st Nile Cataract are unpublished 'rubble' groynes near Karnak in Luxor, perhaps built to facilitate island formation (Bunbury, 2019, p. 109). Notwithstanding the same issues of identification and Nile movement, it seems unlikely that these structures would have been required for agricultural usage in Egypt proper. Basin irrigation techniques (management of water by levees and canals) in use since at least the late Predynastic period already allowed for the efficient irrigation of Egypt's wide floodplains and delta (Butzer, 1976). As such, the need for additional jarf land was probably much less and unlikely to be worth the labour-intensive transportation of building stone.
Nonagricultural uses: Nile navigation and fishing?
Construction for agricultural silt retention cannot account for long, monumental walls that extend well into active channels, where they do not cause significant sedimentation. At Soleb, the longest series of river groynes acts as a partial weir during periods of low river flow, funnelling water through a narrow gap and scouring a deep, fast-flowing channel (Figure 13). Other substantial groynes upstream and downstream of the temple would have increased scour at their tips, similarly contributing to the maintenance of a deep channel entering and exiting this cataract at times of low flow. In such settings, these structures were plausibly built to train the river channel and improve navigation.
Our revised plan of groynes upstream of Soleb temple suggests that they were not intended to block access to the temple's port and create a diversion channel to the east, as previously suggested (Schiff Giorgini et al., 2003, fig. 6a). Conversely, these works would have facilitated navigation through the western side of the main channel. If built during or before the New Kingdom, they would have improved the most direct boat access route to the temple quays (Figure 13). Related, potentially New Kingdom, efforts are evidenced near the Ramesside temple of Gerf Hussein in Lower Nubia, where William Willcocks describes a large, well-built stone wall that redirected Nile flows to scour a deeper harbour in front of the temple (1899, p. 31). Meanwhile, the positioning of the temple of Tiye at Sedeinga, just downstream of a large bedrock spur projecting into the Nile, seems unlikely to be coincidental. Even without augmentation, this feature would have created a sheltered landing place in front of the temple (C. Rilly, personal communication, May 2019). If built during the New Kingdom, the section of wall on the inland side of this spur, comparable in width and stone size to those at Soleb, could have been intended to improve this temple landing. A possible extension of the wall into the channel would have simultaneously assisted in maintaining a deep channel adjacent during periods of low flow. Massive, perpendicular harbour walls are known from several Middle Kingdom forts in Nubia, including Aniba (Vogel, 2004, pp. 219–244) and Buhen (Emery et al., 1979, p. 100). There is no evidence for any such infrastructure at Sai Island (Budka et al., 2020, pp. 58–61) or Amara West, presumably because they were not required due to the nature of local Nile flow (cf. Manzo, 2017).
The walls at Soleb, some 4 m in width, up to 700 m in length and overwhelmingly composed of quartz boulders weighing c. 100 kg, are truly monumental in scale. Further investigation is required to ascertain the volume of stone involved in their construction, but the immense labour involved is nevertheless clear, and probably indicates a state-level initiative. Given the scale of investment and labour required for the construction of Soleb temple and its surrounding earthworks, these structures would obviously have been within reach of its late 18th Dynasty builders.
Intriguingly, civil engineers employed by the Sudanese government to survey the Nile's suitability for long-distance river transport noted 'traces of former barrages constructed to alter water levels' in multiple cataracts between Dongola and the High Dam reservoir (Greenaway, 1986, p. 37).6 This observation reminds us that the walls' position near Soleb need not have related solely to facilitating New Kingdom access to the temple itself. Instead, they could have formed part of a wider programme of river improvements, whether undertaken during periods of pharaonic rule or at other times. The monumental quartz structures identified at Sai Island nearby, of similar construction method and width to those at Soleb, provide further possible examples of river training groynes. To the north, Vercoutter describes multiple comparable structures in the Batn el-Hajar (1966). These include massive 'spur walls' at Semna, interpreted as a means of concentrating river flow and improving navigation, in this case for Middle Kingdom military purposes and for control over the borders of empire. Where the river could not be trained, other solutions were found, such as the monumental portage slipway built near Mirgissa, used to drag boats around a particularly difficult stretch of the 2nd Cataract at low water. This appears to have been in use throughout the Middle Kingdom and perhaps into the New Kingdom (Creasman & Doyle, 2010, pp. 18–20). Inscriptions also attest to state investment in canal cutting at the 1st Cataract to aid navigation in the Middle and New Kingdoms (Quirke, 2009, pp. 223–227).
Upper Nubia lay outside the area controlled by the Middle Kingdom pharaonic state. Once under Egyptian rule from around 1450 B.C.E., the New Kingdom state would certainly have had the motivation to build such structures in the region. River transport through Nubia was of huge importance, not least for exporting resources to Egypt (see above). The transport of heavy cargoes such as building stone and monumental sculptures for temple construction would have also required deep channels. It is also possible that the Kerma state undertook these works before New Kingdom colonisation, not only to facilitate contact and trade with Egypt (cf. Manzo, 2017) but also to improve access to and control over their extensive territories (including an important centre on Sai Island; see below). A later construction date cannot be ruled out. Due to their position in active channels, the independent dating necessary to place these monuments within their historical context will be impossible to undertake unless buried segments can be located.
Owing to their convoluted shape and flimsy structure, the distinctive cellular, drystone features located at Sai Island could not have served as silt traps or river training groynes. Instead, they most closely resemble the stone or wooden fish traps built in marine and riverine contexts across the world, including Australia (e.g., McNiven et al., 2012; Rowland & Ulm, 2011), Ireland (Montgomery et al., 2015) and South Africa (Gribble, 2006). The fully enclosed structures could have trapped fish as the flood receded.7 In others, fish could have entered through narrow openings or excavated channels. Once inside, they would have been far easier to catch. The Nauri Decree of Seti I, a royal text carved into a cliff face downstream of the 3rd Cataract around the time of the foundation of Amara West, refers to 'fishing pools which are in the whole land of Kush (Upper Nubia)' (Edgerton, 1947, p. 222). Further radiometric dating is required to establish a chronology for these features.
Towards a landscape history of Amara West
The positioning and dating of river groynes at Amara West, combined with recent and older survey and excavation data, aid in reconstructing the long-term landscape history of this area and the wider region.
Across the Abri–Kosha reach, very few groynes are located on the seasonally flooded, mainland (northern) side of the northern palaeochannel belt (Figures 3 and 5). This likely relates to the difficulties of undertaking décrue agriculture on the upwind side of these channels, due to the scouring or burial of saluka soils and crops by sand carried on prevailing winds. Indeed, modern farmers in this reach tend to avoid the northern banks of the Nile, except in narrow strips of riverside saluka land protected by tall riparian dunes and tree cover. The main exception to this distribution is a cluster of walls and linear stone scatters situated on gently sloping, former saluka land adjacent to the mainland in site 2-S-44 (Figures 10c,d). These structures may provide evidence for groyne construction before large-scale aeolian sand influx, which, at Amara West, was well entrenched by the late second millennium B.C.E. (Woodward et al., 2017). Alternatively, these groynes could have been built in an area protected from sand by now-lost riparian dunes.
Currently, the earliest possible radiometric date for groynes excavated at Amara West, in the 17th century cal. B.C.E. (wall 1300), falls before the New Kingdom conquest of Upper Nubia. Activity in the environs of Amara West around this time is indicated by ceramic evidence from 2-R-19/2-R-19a (see above) and by sites recently surveyed by the Munich University team (Budka et al., 2019, Map 6). Moreover, there was evidently a significant Kerma community in the region at this time, most clearly represented by the elite cemetery on Sai Island (Gratien, 1986), with sizable stone-capped tumulus tombs attesting to the availability of large-scale organisational and labour resources. As such, it is plausible that the indigenous Nubian inhabitants of the area built this and other walls (as suggested by Vila and Arkell; see above), although caution is merited due to the lack of bracketing date for the construction of wall 1300. At any rate, we should not assume that extensive hydraulic infrastructure in this region must be the work of the pharaonic state.
Evidence for early 18th Dynasty activity in the hinterland of Amara West, contemporaneous with the foundation of and early decades of occupation at the nearby pharaonic town on Sai Island, includes pharaonic seal stamps and Egyptian-style, wheel-thrown pottery, recovered from buildings with mudbrick and/or dry-stone walls at sites 2-R-18 and 2-R-65 (Stevens & Garnett, 2017, pp. 303–305) (Figure 5). Perhaps seasonally occupied, these sites (base camps and associated middens?) may relate to the control of desert routes and the extraction of resources, including gold-bearing quartz and pigment deposits (Fulcher et al., 2022). No directly dated walls fall within this period, but it is plausible that the saluka land associated with any pre-existing structures remained under cultivation and that new walls continued to be built.
At the time of its foundation, the walled town of Amara West would have been situated in the midst of a substantial area of good agricultural potential. According to the HEC-RAS model presented above (Figure 3), 12.3 km2 of land in the Abri–Kosha reach north of the present Nile channel would have been inundated at maximum flood. OSL dating indicates that wall 1343 was constructed between 1205 B.C.E. (date range 1379–1031 B.C.E.) and 700 or 440 B.C.E. (date range 855–305 B.C.E.). Whilst too imprecise to enable correlation with historical events, these dates are partly consistent with construction during the New Kingdom occupation of Amara West (c. 1300–1070 B.C.E.). Furthermore, OSL dating of the Amara West palaeochannel and the northern palaeochannel belt indicates that these permanently dried by the late first millennium B.C.E. (see above). The hundreds of examples located within these palaeochannel systems must have therefore been built before this. Current evidence for the area's settlement history indicates little activity around these channels after c. 1000 B.C.E., suggesting that the walls were most likely constructed during the New Kingdom or earlier periods of intensive inhabitation. It is also most likely that the structures were predominantly built before or around the time that these channels ceased to flow perennially/seasonally around 1270 B.C.E. (date range 1485–1055), when décrue cultivation could have taken place annually.
There is little doubt that Amara West's pharaonic administrators would have had cause to improve the area's agricultural productivity. The interest of the pharaonic state in controlling food production in occupied Nubia, whether for local storage and consumption of produce or for onwards supply to Egypt, is evidenced by large granary and silo facilities in the region's major towns (Spencer, 2019, pp. 441–443, 453), alongside the presence of officials with titles such as 'overseer of the double granary' (Amara West; Spencer et al., 2014, pp. 19, 34) or 'overseer of cattle' (Auenmüller, 2018, p. 251 [6.21]). The latter is relevant here as livestock would have required abundant pasture or fodder. The aforementioned Nauri Decree describes punishments for those interfering with agricultural, fish and livestock resources from Nubia, intended for a royal temple foundation in Egypt (Edgerton, 1947). There is also evidence for differential access to grain supplies between larger and smaller houses at Amara West (Ryan et al., 2012, p. 105), suggesting possible elite control over staple agriculture. While these factors lend weight to a hypothesis that the pharaonic state was responsible for constructing some groynes, it must be borne in mind that ethnoarchaeological analogy suggests that these structures were well within the reach of individuals and small groups, and could thus also have been built on the initiative of nonelite farmers themselves.
As outlined above, the drying of the northern and Amara West palaeochannel systems led to the degradation and ultimate loss of much of this productive landscape. Yet, there is evidence that some channels within the southern palaeochannel belt continued to flow much later. The X-Group/Early Christian site 2-S-37 (Stevens & Garnett, 2017, pp. 288) is located on one or more palaeoislands within this channel belt (Figure 3). The site was probably only habitable because an active channel shielded it from the aeolian sands that have subsequently covered it. Interestingly, 25% of walls in site 2-S-36, all of which are positioned within the southern palaeochannel belt, are built of flat boulders placed upright and perpendicular to the structure. This construction method is almost wholly unrepresented in sites with walls located predominantly (2-R-14) or wholly (2-S-44) in the northern palaeochannel belt, in spite of identical construction materials. Drystone walls of upright slabs are common in nearby Christian period monuments such as Girgetti (Vila 1977b, p. 35, fig. 12) and Soumbout (Vila, 1978, pp. 29–30, fig. 4), and this building method was also used in 20th century C.E. groynes on Sermutta and in (probably) similarly recent examples at Sai Island and Quweyka adjacent. This suggests a more recent vernacular construction method, although further direct dating is required to prove this.
5 CONCLUSION
The study of ancient Nubia has been shaped by ancient and modern colonialism, with a focus on monumental temples, walled towns and elite cemeteries, particularly those created under successive periods of pharaonic rule. These studies have often emphasised similarities with contemporaneous features in Egypt proper (e.g., Edwards, 2004; Spencer et al., 2017). Yet, the floodplains in this region, particularly in cataract reaches, are far narrower than those of Upper Egypt. To sustain large communities here would have required agricultural strategies different to those of Egypt, of which the deployment of river groynes that we describe here was likely a key element. These structures probably contributed significantly towards the agricultural productivity of large parts of this region, especially before the introduction of the sagia in the 3rd century C.E.
This study has demonstrated that dating ancient river groynes is possible, albeit challenging. In the environs of Amara West, radiometric dating of wall-associated deposits and ecofacts provides earliest possible dates for construction ranging from the mid-second to late first millennia B.C.E. and latest possible dates ranging from the early to late first millennium B.C.E. Additionally, OSL dating of the Holocene riverine landscape surrounding the site suggests that many examples were built during or before the late second millennium B.C.E. Elsewhere, robust chronologies of palaeochannel belts in the Dongola Reach and stratigraphic relationships from Wawa suggest groyne construction before c. 280 C.E. (and most likely before c. 1290 B.C.E.) and the first half of the first millennium C.E., respectively. Some groynes in this region thus pre-date the earliest known examples of these structures anywhere in the world by over two millennia. Their construction in northern Sudan in recent centuries is evidenced by multiple ethnographic sources, indicating a riverine tradition of remarkable longevity.8 It is possible that river groynes were in use here much earlier than indicated by current dating. Silt will bank up around natural rock spurs during the flood, and this phenomenon may have been noted and emulated by different groups at different times. Indeed, the region's inhabitants would have had good reason to increase the amount of naturally irrigated saluka land for as long as cereal agriculture was practiced in this stretch of the Nile Valley, from the turn of the fourth millennium B.C.E.
Further investigation is necessary to understand the genesis of this form of channel management and its use over time. In particular, it is essential to deploy further radiometric dating to tie the practice into broader historical trends in demography and human–landscape interactions and to better establish the hydroclimatic and fluvial geomorphological contexts of both these structures' construction and the abandonment of associated agricultural land. Phases of groyne building may mark shifts to new flood regimes, whilst abandonments may relate to episodes of river channel contraction that have been linked to climate-driven falls in Nile flood magnitude (Macklin et al., 2013, 2015; Woodward et al., 2015, 2017). These hypotheses require further analysis.
In contrast to these structures, the distinct class of monumental river groynes recorded at Soleb, Sai and in the now-flooded Batn el-Hajr and Lower Nubia was probably built to train the Nile for improved navigation at times of low flow. At present, their construction can be only tenuously dated by association with nearby monuments. These were most likely state-level initiatives, in many cases undertaken by the pharaonic state and, in Upper Nubia, perhaps by Kerma. Other monumental groynes facilitated the harbouring or landing of rivercraft.
The control, manipulation and transformation of the Nile, facilitating the movement of armies, communities, individuals, ideas, technologies, foodstuffs and raw materials—and enhancing agricultural productivity and economic opportunities—would remain a focus for states and imperial ventures into the modern era (Tvedt, 2004). The data presented here also remind us of the role played by local communities in transforming riverine landscapes.
ACKNOWLEDGEMENTS
This research was funded by a generous grant to M. Dalton from the Michela Schiff Giorgini Foundation (prix à la mémoire de Jean Leclant 2018). Fieldwork was undertaken within the framework of the British Museum's Amara West Research Project (directed by N. Spencer), funded by the Qatar Sudan Archaeological Project (Grant A.007, 2013–2019). The work builds upon research undertaken as part of the Leverhulme Trust-funded project Health and diet in ancient Nubia through political and climate change (F/00 052/C; 2010–2014) and the Australian Research Council-funded project Environmental impacts of climate change in the Nile Basin over the past 30,000 years (ARC DP0878058). We are grateful to the National Corporation for Antiquities and Museums, Sudan, for providing the archaeological fieldwork permit, authorising the export of samples for scientific analysis, and for drone hire. Special thanks are due to Mohamed Saad (NCAM inspector) and Reham Mohammed (assistant NCAM inspector) for their fieldwork contributions. Finally, our thanks to David Edwards for providing aerial photographs of the Batn el-Hajar, Mohammed Hassan for translating during ethnoarchaeological interviews and Vincent Francigny and Claude Rilly for facilitating the examination of walls in SFDAS concessions (Sai, Soleb and Sedeinga) and for feedback on draft results. We acknowledge two peer reviewers for their constructive comments. Open access publishing was facilitated by The University of Western Australia, as part of the Wiley - The University of Western Australia agreement via the Council of Australian University Librarians.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflict of interest.
-- Sent from my Linux system.
No comments:
Post a Comment