Linear hollows in the Jazira, Upper Mesopotamia.

Wilkinson, T.J.

O.G.S. Crawford, founder of ANTIQUITY, flew in the 1920s over an English landscape where the grooves and lines cut into unploughed downlands showed the courses of roads and tracks since earliest times. Similar patterns of crop- and soil-marks in the rain-fed agricultural zone of the Middle East, when studied in the same spirit, also reveal the local and the long-distance routes of a proven great age.

Introduction

Expansion and intensification of archaeological surveys during recent years has opened up the entire landscape for investigation. To complement the myopic focus upon settlement sites, techniques of off-site archaeology have started to unveil traces of past land-use systems, water supply, communications, quarrying and other landscape features. Such studies are not new, for they were pioneered, as field archaeology, earlier this century especially when aerial archaeology was developing as a key archaeological tool (Beazeley 1919; Crawford 1923; 1953). Today, air photographs and satellite images are even more valuable, partly because techniques of off-site archaeology and survey supply more detailed ground data to act as control for remote sensing studies. Features once interpreted by assumption or on rather flimsy evidence can now be examined critically.

The features here entitled 'linear hollows' are particularly evident over thousands of sq. km of the dry farming zone of the Jazira of northern Syria and Iraq. In a classic paper Van Liere & Lauffray (1954) described, mapped and interpreted these features as ancient routes hollowed out of the landscape by the sustained passage of humans and animals (similar observations in Crawford 1953: 8; Buringh 1960: 212-13; Oates 1968: plate 1a; Oates & Oates 1990: plate 66). Related features in the British countryside, sometimes found along existing roads, have long been called 'sunken lanes' or 'hollow ways' (Taylor 1979; Hindle 1982: 11). Other studies have indicated a relationship between similar hollow way routes and prehistoric funerary monuments in the Netherlands (Jager 1985), and an impressive complex of linear concave roads, dating to 900-1150 AD, has been recorded around Chaco Canyon, New Mexico (Obernauf 1980; Powers 1984: 52-3; Gabriel 1991; Warburton & Graves 1992: 57).

As roads, the Jazira features are fundamental to the ancient geography of this region which can then be divided into settlement sites (tells and other sites); systems of intensive land-use and manuring (represented by some off-site artefact scatters: Wilkinson 1989); and communications (the linear hollows). This paper describes linear hollows, suggests mechanisms of formation and discusses alternative interpretations that have been proposed. Specifically, it has recently been suggested that such features in the Jazira may have been canals, or at least conduits for channelling run-off to fields to enhance soil moisture in this marginal environment (McClellan n.d.). As will be shown below, such a conclusion is unwarranted, given the relationship of the features to topography as well as the absence of cut channels, upcast or clean-out silts.

The Jazira and the selected case studies

The undulating plateau of the Jazira, developed on Tertiary sedimentary formations, generally fluctuates between 300 and 450 m above sea level, except where broken by upstanding anticlinal hill masses such as the Jebels Abd al-Aziz and Simjar (FIGURE 1). The name Jazira (Arabic: island) derives from its position between the Tigris and Euphrates rivers. Today its landscape is a highly degraded steppe (Bottema 1989). The northern part receives sufficient rainfall for crops of cereals and legumes in most years; where rainfall is less than about 250 mm per annum, crops fail too frequently to be economic. The zone of potential dry-farming has fluctuated with social, political and environmental conditions. When political control was weak during the late Ottoman period, the limit of cultivation was in the moister northern steppe; during the 20th century, with more political control and investment, settlement and cultivation spread to an extreme southern limit close to the 200-mm isohyet (FIGURE 1) where farming is very risky (Lewis 1955).

The three case study-areas are: Tell Sweyhat in Syria (rainfall 200--300 mm p.a.), the North Jazira Project, Iraq (300--400 mm), and Kurban Hoyuk in southeast Turkey (400--500 mm) (FIGURE 1, which also shows the central Jaziran steppe region of Van Liere & Lauffray's study).

[CHART OMITTED]

Tell Sweyhat, Syria

The area of Tell Sweyhat on the east (left) bank of the Euphrates in northern Syria, shows partial development of linear hollows in an area of moderate topographic complexity (for excavations at Sweyhat see Holland 1976; 1977; for near-by Tell Hadidi, Dornemann 1988). In what is now part of Lake Assad, the Euphrates river is entrenched some 20--30 m into Pleistocene terraces, or, in the south, low hills of Tertiary limestone (FIGURE 2). With the exception of the Euphrates, the drainage is a dendritic net of slightly sinuous ephemeral wadis. Wadi erosion is enhanced near the Euphrates where scour by the migrating channel removes the alluvial/colluvial fans and lower reaches of tributaries so that the wadis incise back from the precipitous cut. This increase in wadi gradient and flow power rejuvenates the wadi and a 'knick-point' migrates upslope. Further away from the river, most wadis have gentle cross-sections, and any incision is due to local catchment characteristics. here the wadis either have active aggrading gravel-bed channels, or have become infilled by millennia of plough wash. Where cuts exist, they suggest that the wadis have a sedimentary history extending back through the Holocene to a time before sedentary human settlement.

A number of straight swales or linear hollows are apparent on air photographs. Although their straightness distinguishes them from wadis, it is clear from FIGURE 2 that some at least form part of the natural drainage. Most hollows near Sweyhat have been seen on the ground, but the more isolated examples and those to the west of the river (possibly now flooded by the waters of Lake Assad) have not. In the field, they vary from prominent topographic valleys (e.g. FIGURE 2, b) to mere swales having virtually no surface expression. Sometimes the only indication on the air photograph is a slight vegetation mark (FIGURE 2, e; cf. Oates 1968: plate 1a). The linear hollows can be seen to radiate from, or to run between, sites. An alignment leads up the Euphrates to the northwest of Sweyhat (FIGURE 2, a). A pair of swales aligned to the southwest of Sweyhat lead via a deep straight valley towards Early and Middle Bronze Age Tell Juaf (or Jouweif; FIGURE 2, b & c). In the vicinity, the southern of the two swales (c), although subtle, can be seen to lead towards a gap in the outer wall that is a suspected gate. To the east of Tell Sweyhat hollows cannot be recognized because the air photo coverage terminates immediately to the east; here, localized sheet sedimentation may have obscured any subtle topographic features. Three features radiate to the north of the Early, Middle and Bronze Age centre of Tell Hadidi. Other alignments are focussed upon Tell Othman (d) or lead to the southeast from Tell Juaf (e & f).

[CHART OMITTED]

In some cases the hollows actually form part of the trunk or main wadi channel; elsewhere they only include part of a minor tributary. Occasionally they act as a focus for gully erosion, and wadi channels develop along part of them (e.g. FIGURE 2, a, b, c & e ). Their hollow form and their relationship to the drainage net shows they conduct some run-off, but their general discordance with the natural drainage system also shows other factors have influenced thier alignment and formation. Unlike canals, none show evidence of having been excavated, nor are there any signs of banks of upcast or scatters of freshwater molluscs (indicators of perennial flow) alongside. Some (such as b & c) clearly run from one wadi system to another; they have reaches of reversed gradient and cross watersheds. Although the general slope of the terrain (and of the hollows) is from east to west, there is no sign of a water source to the east; indeed the water table drops relative to the ground surface to the east. Nor do they visibly conduct storm run-off on to fields; rather they would have conducted water towards the seasonally moist and locally irrigable flood plain of the Euphrates (FIGURE 2).

The North Jazira Iraq

This small basin of about 750 sq. km total area is some distance from the nearest river, and there has been limited Holocene sedimentary aggradation. Apparently a flat silt/clay plain, the land does have a continuous slope, and after prolonged intense rain, discharges significant run-off. In addition to low-sinuosity 'wadi ridges', probably of early Holocene date, wadis are either very sinuous or are suspiciously straight.

The plain is dominated by the bulk of Tell al-Hawa, probably the Bronze Age regional capital, which was excavated by Warwick Ball for the British Archaeological Expedition to Iraq in 1987 and 1988 (Ball et al. 1989; Ball 1990).

On air photographs the linear hollows comprise:

1 Broad shallow 'swales', that usually radiate from major tells, most of which reached their maximum size approximately during the mid 3rd millennium or slightly later (Wilkinson 1990b).

2 Features that cross the landscape over longer distances or which link the more recent archaeological sites.

Van Liere & Lauffray (1954: 145--6) saw the broad features as radial and Bronze Age in date, and the narrower cross-country features as post-Hellenistic. In the North Jazira, there is a general continuum of features, many being of intermediate size and indeterminate date. The best developed radial patterns are around large tells, the Bronze Age centres. Many hollows are subtle shallow swales 30--60 m in width (but sometimes as much as 200 m), and only 0.5 to 1.0 m deep; others are of the scale of natural drainage channels. At their most subtle they have no surface expression; they show as dark soil marks, which probably result from hollows being infilled with soil wash. Under favourable circumstances they form crop marks or vegetation lines. The cross-country features appear to have been less common in the Syrian Jazira, although a quick inspection of map 1 in Van Liere & Lauffray (1954) reveals occasional long-distance lines.

Radial hollows

Within the North Jazira the developed radial systems focus upon the major Early Bronze Age centres of Tell al-Hawa, Kharaba Tibn and Abu Kula (FIGURES 3 & 4) and other major sites thought to be of similar date, such as Abu Winni (FIGURES 3 & 5) and Abu Wajnam (FIGURE 3). Where measurable, the radial hollows have a modal length of 2--2.9 km, and a mean length of 3 km (the mean length for similar features in the Syrian Jazira is 3.9 km, where there were about 5.4 lines per site: McClellan pers comm. 1991). Twelve hollow ways bifurcate into two at locations 1--2 km from the central site. Around the 66-ha site of Tell al-Hawa, a total of 14 radial lines can be distinguished (FIGURE 4). Most are on the north side of the tell, where the topography is steeper, and the hollows can be remarkably deep and wide (FIGURE 6).

[CHART OMITTED]

FIGURE 4 illustrates the relationship of linear hollows (plotted from air photographs) to topography, whereas FIGURE 5 is from air photographs alone. Around Tell al-hawa they cross the contours at angles of 45[degrees] or more and in all cases converge upon the tell. These characteristics, together with their straightness, distinguish them from the moderately sinuous wadi traces. At Kharaba Tibn and Abu Winni the hollows form a virtually complete radial system around the central tell.

The discordance of the hollows to the general surface slope results in wadi courses flowing along large valleys, and then adopting oblique alignments for short distances until a course approximating to the original is again taken; this is particularly evident at Site 30 and in the vicinity of section C on FIGURE 4. Note how, to the west of Tell al-Hawa (i.e. southwest of C) there is a partial oblique alignment and the wadi flow can follow alternative paths to the south or southeast. Similar coincidence between wadi courses and linear hollows is evident to the west and southwest of Tell Abu Winni on FIGURE 5. Although the hollows conduct episodic flow most are discordant to the usually dendritic pattern of overland flow, and thus show only a partial adjustment to the (developing) drainage system.

As topographic profiles along selected hollows demonstrate, most have a consistent down-slope gradient; but several cross the terrain virtually independent of topography. For example within 2 km of Kharaba Tibn (Site 43, FIGURE 6) profiles I and II show how the linear hollow rises over a watershed at c. 1.5 km from the tell. Traces of the hollow then disappear within the wadi (at 2 km on FIGURE 6), but re-appear to the northwest. Within the catchment of Abu Winni it can be inferred that three hollows to the southwest and west of Abu Winni must rise and fall over several watersheds. Where the hollows cross watersheds, run-off must diverge to follow the gradient, as indicated by the arrows on FIGURE 6.

Long-distance hollows

Among the dense network of features indicated on FIGURE 3 can be distinguished at least 11 longer, albeit discontinuous features. Like the radial features, these sometimes cross watersheds; a long hollow runs to the northeast of Tell al-Hawa towards Hamad Agha Saghir which, according to 1:100,000 topographic maps, negotiates a major watershed some 8 km northeast of Hawa. The best developed long hollow-ways on FIGURE 3 are:

a To the north of Tell Huqna and Abu Winni, thence through Uwaynat where one branch runs through Tell al-Hawa towards Bir Uqla and the northwest, and a second branches slightly to the south to follow the course of the Wadi al-Murr to Tell al-Samir and the northwest.

b To the south of jabal al-Qusair, through Tell al-Dhaim and al-Kibar and al-Gana and thence to the northwest. Other features follow south-southeast--north-northwest alignments (in the southwest near al-Mumi) as well as from southwest--northeast, as from the large but undated site at x along the southern border.

The association of hollows with numerous archaeological sites suggests that they were developing or were in use when such sites were occupied, and thus can be roughly dated. The radial systems focus upon large tells, most of which have histories extending back to the 3rd millennium BC, if not earlier. The branching long-distance feature through Uwaynat and Tell al-Hawa on FIGURE 3 has numerous sites of the 3rd and early 2nd millennium along it (i.e. contemporary with the Akkadian and Old Assyrian periods). An earlier date is hinted at by the presence of Late Uruk sites of southern Mesopotamian type along the southern branch of (a) through Tell al-Samir and the al-Gana alignment (b). It is even possible to suggest that the al-Gana branch formed part of the Emar itinerary referred to in Old Babylonian texts (Wilkinson 1990b: 61) On the other hand numerous Parthian, Sasanian and Islamic sites along the features are later associations which suggests that they have a long time range extending over some 5000--6000 years.

Kurban Hoyuk area, southeast Turkey

Elsewhere in the Jazira, in southeast Turkey to the north of Urfa, linear hollows are present, but networks cannot be mapped because air photographs are not available. The best developed hollow can be traced along a traditional dirt track which runs parallel to the Euphrates (FIGURE 7) on a line separating two types of modern fields: strip fields to the south and block fields to the north (Wilkinson 1990a: figures A4, A15 & A16). Where the track crosses tributaries of the Euphrates it occupies the base of opposing pairs of valleys up to 3--4 m deep (FIGURE 7). The track, and associated intermittent hollow, links a number of Late Roman/Early Byzantine sites. A large limestone column in the bed of the Incesu Deresi (a south bank tributary of the Euphrates, FIGURE 7) is located mid-way between an opposing pair of linear hollows along the line of the traditional track. Originally interpreted as a bridge pier (Wilkinson 1990a: 119), it is more likely a Roman milestone without an inscription (cf. an example from Deir Sam'an, Samaria: Dar 1986: plate 13). In this case the

association of the hollow with a traditional trackway and a Roman milestone provides a good case for this landscape feature being of at least Late Roman date; it was then perpetuated through continued later use.

[CHART OMITTED]

Discussion

Mechanisms of formation

Both the form of the linear hollows and their place in the natural drainage net suggests that they do in part have a hydraulic function, but the absence of diagnostic features of canals (described above) argue against them having been cut to conduct irrigation water.

The movement of large numbers of humans and animals along tracks will tend to compress soils, thus decreasing infiltration and increasing the rate of run-off (Frenkel 1970, in Goudie 1986: 61). By concentrating flow and increasing its power, both erosion and sediment transport along selected paths is encouraged. The churning action of feet and hooves will destroy soil structure and markedly increase sediment yield. Wheeled vehicles, in use from as early as the 4th millennium BC (Littauer & Crouwel 1979: 13), cause ruts which concentrate flow and increase flow power and eroison. The occasional presence of two contiguous hollows (as near C on FIGURE 4) can be explained by the abandonment of an earlier unusable track and the adoption of a new path, constrained by the enclosed fields on either side.

Selective and concentrated overland flow along a generally straight alignment explains the tendency of linear hollows to cross watersheds without the need to invoke canals with associated engineering works. The line of the track provides the zone of maximum erosion and sediment movement, but within each fluvial catchment water will flow downslope and away from the watershed.

During the long dry summers, flocks and herds moving along routes will churn up dust which blows away. Apparently negligible, this will have a significant effect over thousands of years. Just 1 mm of dust blown away per year (equivalent to 10--15 tons per ha; Mainguet 1991: 194) lowers the land surface by 3 m over 3000 years. This in turn concentrates overland flow, leading to further erosion along the selected route. Although interaction between runoff and wind erosion is probably considerable, the relative contribution of the two agencies remains an open question.

In the American Southwest, clearance of stones and soil is thought to contribute to the concave profile of similar features (Obernauf 1980: 139); this may have contributed to the hollowing in the stonier parts of the Jazira. Certainly the shovelling away of mud has helped create 'hollow ways' in Britain, where manorial records indicate that manure that accumulated along roads was often shovelled on to near by fields (Tom Williamson pers. comm. 1990; Taylor 1979: 145--6). Clearance of stones and surface rubble was also crucial to road construction and maintenance in the Samaria region, Palestine, where road and field path surfaces were cleared down to the bedrock below. Many such tracks, which date back to at least the Hellenistic period, also had high 'kerbs' of stones alongside, demarcating the division between the road and adjacent fields and also keeping animals out of the fields (Dar 1986: 126--7).

Studies of arroyas (deep stream gullies) in the western United States have shown that the effects of flow concentration factors -- cattle trails, roads, embankments and indeed canals -- all serve to increase stream power locally and lead to gully incision (Cooke & Reeves 1976: 178--9). Although the subtle features of the Jazira show few traces of gully incision, where gradient and conditions for sediment evacuation are suitable, near Tell es Sweyhat (FIGURE 2) and Abu Winni (FIGURE 5), the relationship between episodes of gully erosion and human-induced flow concentration features is evident.

A central tell on a flat or gently sloping plain would be served by a radiating net of routes. Such a pattern is evident around present-day villages in the Jazira and is illustrated by the radial system of field tracks and village roads that developed in Palestine during the 1st millennium BC (Dar 1986: figure 83). Although this model fits some of those in FIGURE 3, some explanation is required where only part of a radial net remains, as in the vicinity of Tell al-Hawa.

How may interaction between a route pattern and natural processes of run-off, erosion and deposition influence valley development? Immediately northwest of C near Tell al-Hawa (FIGURE 4), the valley that probably originally flowed orthogonal to the contours directly south across the plain, now flows to the southeast towards the tell. To the south of Tell al-Hawa a lobate feature, evident in the contour lines at and beneath the Hellenistic and Islamic Site 6, may be an alluvial fan where sediments eroding from the radial valleys to the north were deposited. Here a pre-Hellenistic date would not conflict with the erosion of the hollows which probably occurred during the period of peak size and population at Tell al-Hawa during the early Bronze Age. Although the above association is tenuous, a complex history of Holocene erosional channel changes is suggested by the presence of in-filled wadi channels containing Neolithic Hassuna and Akkadian pottery (i.e. 6th and 3rd millennium BC) to the west and south of Tell al-Hawa (FIGURE 4, at x and south of y). Thus around Tell al-Hawa, a northern zone of erosion of linear hollows and valley development is complemented by a southern zone of aggradation. The depositional products of this erosion, infilling older valleys, may have obscured earlier linear hollows.

An additional factor is the size of the run-off catchment (FIGURE 8). If a tell with radial routes develops on land of gentle slope (similar to Tell al-Hawa), the catchment size of each radial route (flow concentration zone) varies, with the upslope catchments receiving the highest proportion.

[CHART OMITTED]

In the North Jazira where long linear hollows traverse the terrain, there has been considerable interaction between hollows and wadi drainage. Normally channels in this silt/clay plain are sinuous, but the occasional long, straight reaches through areas with intermittent traces of linear hollows (e.g. Uwaynat to Tell al-Samir on FIGURE 3) suggest that wadi flow has adopted selected hollows during phases of flooding.

In the Khabur basin of Syria radial routes focused upon multi-period Bronze Age tells were wide and deep. Conversely, long-distant routes were narrower and associated with sites of Hellenistic or later date (Van Liere & Lauffray 1954: 146). The width and depth of hollows may be proportional to the age of the associated sites and their length of occupation (Van Liere & Lauffray 1954: 146). In the North Jazira, radial routes were mainly broad and deep, whereas cross-country routes were of all sizes. The presence of Bronze Age sites along many cross-country routes suggests a history extending back some 5000 years. Run-off conditions being equal, the smaller features probably carried less traffic, were in use for less time, conducted more wheeled vehicles or were later features.

Traffic flow can now be estimated around central tells and along cross-country features. Although radial systems (FIGURES 3, 4 & 5) occasionally link satellite settlements to the central tell, many die out between 2 and 4 km. Such hollows could be field tracks that conducted inhabitants from villages to outlying fields, a common feature for agricultural communities (Dar 1986: 133). In the Jazira, flocks and herds would also have used such routes to reach grazing on fallow fields or outlying pastures. Nominal figures can be suggested for movement along a single track from a medium size tell of 20 ha area, capable of housing some 2000--3000 people, half of whom were involved in agriculture. Average household size is assumed to be 6, and each site has six radial tracks.

Cultivated zone (per route): 200 field workers working 25 days per month for 6 months, ploughing, manuring, weeding, harvesting, gathering in the crop, etc. (to include at least 30 draft animals): 34,500 person/animal movements x 2 = 69,000 total journeys, assuming that the same route is used for the return journey. A lower figure results if hours worked per unit area are scaled up for a 4-km radius catchment (i.e. c. 5000 ha) biennially fallowed (basic data derived from Russell 1988: table 41). This indicates that 16,700--20,800 single trips (33,000--42,000 total journeys) would be required. In addition, to haul to a central place the grain yield would entail in excess of 2000 return ass trips per route, or one-fifth of that number if carts were used (load carrying capacities of 75 kg for an ass, 150 for a mule and up to 500 kg for wheeled carts, are assumed, see Dar 1986: 145).

Pastoral activity: 35 household flocks of 20 sheep (Cribb 1991: 36) with 3 shepherds, either returning along the same route or returning along a different route so that another flock will return along their own outward route: 294,000 single trips or 588,000 return trips. No allowance is made for palace or temple flocks, although these are known to have been large (Gelb 1986: 158).

Unless flocks were taken beyond the area for extended periods, the local route systems would suffer significant wear, with the pastoral use accounting for most of this, especially during wet winters. Inter-regional traffic or movement between centres would need to approach this figure of some 600,000 annual movements to produce an equivalent mark on the landscape. Because many tells were occupied for several thousand years, the magnitude of some of the radial hollows is not surprising. By comparison, in semi-arid parts of Africa, significant zones of intensive trampling, devegetation and other degradation can extend up to 3 km from wells within only a few years (Mainguet 1991: 93 & 109).

The distribution of systems of linear hollows corresponds approximately to terrain that can be farmed without recourse to irrigation. Further south, along the valleys of the Euphrates and Khabur, a number of genuine irrigation canals can be traced in part by linear banks of upcast (Ergenzinger et al. 1988). But why is this pattern of hollows so well developed in the region indicated on FIGURE 1, but less clear elsewhere? This can partly be explained by the patchy availability of air photographs, the most spectacular systems having been mapped by Van Liere & Lauffray in an area that was once well covered by aerial photography. Nevertheless, the available data suggests that the features are most strongly developed between the rainfall isohyets of 250 and 400 mm (FIGURE 1).

South of the southern limit of rain-fed cultivation, linear hollows are less common and radial systems are scarce or absent, an exceptional case being the radial lines around Hatra (Bradford 1957: plate 24; Jabar Khalil Ibrahim 1986: plate 56). There is a similar decrease in the clarity of the features northward into the moist steppe. In the Assyrian plains east and north of Mosul, modern fields are small, perhaps due to the operation of divided inheritance over long periods. This area of greater population density and 'resilience' (Oates 1968: 16--18) remained occupied when more marginal parts of the Jazira were abandoned. The scarce and indistinct linear hollows there may result from continued cultivation over long periods which caused the infilling and blurring of hollows. Continuity of occupation would also have led to roads following the same course over long periods in the way that British sunken lanes become enshrined in the landscape and do not become separate features. In contrast, marginal areas of the Jazira with rainfall between 250 and 400 mm per annum suffer episodic abandonment which result in discontinuity in landscape features. In the North Jazira this is illustrated (for Ottoman period abandonment) by air photographs that reveal a more archaeologically pristine landscape than in the moister zone.

The zone of 250--400 mm annual rainfall also conforms to the zone of maximum erosion and sediment yield. In areas with <200 mm p.a. rainfall, although the protective cover of vegetation is low, rainfall and net sediment transport rates are also low. These rise as annual rainfall increases further north (within the 250--400 mm zone) where vegetation cover remains insufficient to form a protective cover, and then declines in still moister areas where vegetation cover restricts surface erosion (see Langbein & Schumm 1958; Mabbutt 1977: 68--9; Mainguet 1991: 194--5). Although the response of areas of linear hollows will be more complex than this model suggests, owing to trampling, devegetation, ploughing, cropping and other factors, the fundamentals of climate and run-off probably at least partly determine their formation and preservation.

Linear hollows and land-use

FIGURE 9 indicates that intensively cultivated land around early Bronze Age centres falls consistently within the mean length of radial hollows. A restricted length for routes is characteristic of post-medieval British villages where tracks went out to the fields but rarely extended even as far as the village boundary (Hindle 1989: 21). Similar radial routes within the relict Iron Age to Roman landscape of Samaria (West Bank Palestine) extended approximately to the assumed territorial boundary. However, in the case of two contiguous settlements there is an impression of a slight overlap in reconstructed territorial boundaries (Grossman & Safrai 1980: 449; Dar 1986: 135). An example from Woolbury Fields in southern England illustrates how hollow ways within an area of enclosed 'Celtic fields' fan out onto a prehistoric grazing ground (Crawford & Keiller 1928: 154). The resultant dispersal of flocks and herds spreads the effects of ground disturbance and reduces flow concentration which in turn inhibit the development of hollow ways. The above cases demonstrate that radial routes can be used either as a proxy limit for enclosed cultivation or for the territorial limit of sites.

[CHART OMITTED]

By this principle, it is possible to model the agricultural territory surrounding a central tell by combining the evidence for ancient land-use intensity with that of radial hollows. Off-site sherd scatters, a proxy indicator of land-use intensity, are assumed to be proportional to the quantity of settlement-derived organic waste applied to the near-by fields as fertilizer (Wilkinson 1989). In the case of the North Jazira, the combined data sets provide the following land use zonation:

a An inner intensively manured and cultivated zone with dense off-site sherd scatters spreading to 1--2 km from the central tell.

b An intermediate zone between 2--3 km with moderately dense sherd scatters and where agriculture was presumably at an intermediate level of intensity.

c An outer zone where sherd scatters, although present, are sparse and have no discernible trend. This is the zone where most radial hollows cease to be evident. The sparse sherd scatters imply a low intensity of manuring, probably with long fallowing intervals.

In the area of Tell al-Hawa, the putative territorial limit lay slightly outside the ring of Bronze Age satellite communities; beyond this boundary the land would either have been very low intensity cultivation with long fallow intervals, or long-term pasture.

If such a relationship between land-use zones and radial routes operated in other parts of the Jazira and sedimentary aggradation has been minimal, it follows that Van Liere & Lauffray's map with its numerous stellar configurations of route systems must also give an impression of the cultivated zones associated with Bronze Age settlement (FIGURE 9, below). This reiterates the point made initially: landscape features should not be viewed in isolation, but in concert with other types of evidence, so that more comprehensive and plausible models can be synthesized than are possible using single data sources in isolation.

Acknowledgements. The results described here derive from fieldwork conducted as part of the following projects: The Oriental Institute project at Kurban Hoyuk, Turkey, the joint Pennsylvania-Oriental Institute Project at Tell es-Sweyhat, Syria, and the British Archaeological Expedition to Iraq project in the North Jazira. I wish to thank the British School of Archaeology in Iraq, for support during field work in Syria and Iraq; Leon Marfoe, Thomas Holland, Richard Zettler and Warwick Ball for help and advice in the field. I particularly wish to thank the following for discussing aspects of the problem of linear hollows: Thomas McClellan, Joan and David Oates, Don Whitcomb, Rodger Grayson and Tom Williamson.

References

BALL, W. 1990. The Tell al-Hawa Project: the second and third seasons of excavations at Tell al-Hawa, 1987--88, Mediterranean Archaeology 3: 75--92.

BALL, W., D. TUCKER & T.J. WILKINSON. 1989. The Tell al-Hawa Project: archaeological investigations in the North Jazira 1986--87, Iraq 51: 1--66.

BEAZELEY, G.A. 1919. Air photography in archaeology, Geographical Journal 53: 330--35.

BOTTEMA, S. 1989. Notes on the prehistoric environment of the Syrian Djezireh, in O.M.C. Haex et al. (ed.), To the Euphrates and beyond: archaeological studies in honour of Maurits van Loon: 1--16. Rotterdam: A.A. Balkema.

BRADFORD, J. 1957. Ancient landscapes: studies in field archaeology. London: Bell & Hyman.

BURINGH, P. 1960. Soils and soil conditions in Iraq. Baghdad: Republic of Iraq, Ministry of Agriculture.

COOKE, R.U. & R.W. REEVES. 1976. Arroyas and environmental change in the American Southwest. Oxford: Clarendon Press.

CRAWFORD, O.G.S. 1923. Air survey and archaeology, Geographical Journal 61: 342--66. 1953. Archaeology in the field. London: J.M. Dent.

CRAWFORD, O.G.S. & A. KEILLER. 1928. Wessex from the air. Oxford: Clarendon Press.

CRIBB, R. 1991. Nomads in archaeology. Cambridge: Cambridge University Press.

DAR, S. 1986. Landscape and pattern. Oxford: British Archaeological Reports. International series 308.

DORNEMANN, R.H. 1988. Tell Hadidi: one Bronze Age site among many in the Tabqa Dam salvage area, Bulletin of the American Schools of Oriental Research 270: 13--42.

ERGENZINGER, P.J., W. FREY, H. KUHNE & H. KURSCHRER. 1988. The reconstruction of environment, irrigation and development of settlement in the Habur in north-east Syria, in J.L. Bintliff, D.A. Davidson & E. Grant (ed.), Conceptual issues in environmental archaeology: 108--28. Edinburgh: Edinburgh University Press.

FRENKEL, R.E. 1970. Ruderal vegetation along some California roadsides. Berkeley (CA): University of California Press. University of California Publications in Geography 20.

GABRIEL, K. 1991. Roads to Center Place: a cultural atlas of Chaco Canyon and the Anasazi. Boulder (CO): Johnson Books.

GELB, I. 1986. Ebla and Lagash: environmental contrast, in H.Weiss (ed.), The origins of cities in dry farming Syria: 157--67. Guilford (CT): Four Corners.

GOUDIE, A. 1986. The human impact on the natural environment. Cambridge (MA): MIT Press.

GROSSMAN, D. & Z. SAFRAI. 1980. Satellite settlement in western Samaria, Geographical Review 70: 446--61.

HINDLE, B.D. 1989. Medieval roads. Princes Risborough: Shire.

HOLLAND, T.A. 1976. Preliminary report on excavations at Tell es-Sweyhat, Syria 1973--4. Levant 8: 36--70. 1977. Preliminary report on excavations at Tell es-Sweyhat, Syria, 1975, Levant 9: 36--65.

IBRAHIM, JABIR KHALIL. 1986 Pre-Islamic settlement in Jazira. Baghdad: Republic of Iraq, Ministry of Culture and Information, State Organisation of Antiquities & Heritage.

JAGER, S.W. 1985 A prehistoric route and ancient cart-tracks in the Gemeente of Anloo (province of Drenthe), Palaeohistoria 27: 185--202.

LANGBEIN, W.B. & S.A. SCHUMM. 1958. Yield of sediment in relation to mean annual precipitation, Transactions of the American Geophysical Union 39: 1076--84.

LEWIS, N. 1955. The frontier of settlement in Syria, 1800--1905, International Affaris 31: 48--60.

LITTAUER, M.A. & J.H. CROUWEL. 1979. Wheeled vehicles and ridden animals in the Ancient Near East. Leiden: E.J. Brill.

MAINGUET, M. 1991. Desertification: natural background and human mismanagement. Berlin: Springer Verlag.

MABBUTT, J.A. 1977. Desert landforms. Cambridge (MA): MIT Press.

MCCLELLAN, T. N.d. Irrigation and hydraulic systems in Syria. Unpublished manuscript.

OATES, D. 1968. Studies in the ancient history of Northern Iraq. London: British Academy.

OATES, D. & J. 1990. Aspects of Hellenistic and Roman settlement in the Khabur Basin, in P. Matthiae, M. Van Loon & H. Weiss (ed.), Resurrecting the past: a joint tribute to Adnan Bounni. Istanbul: Nederlands Historisch-Archaeologisch Instituut.

OBERNAUF, M.S. 1980. A history of research on the Chacoan roadway system, in T.R. Lyons & F.J. Mathien (ed.), Cultural resources remote sensing: 123--67. Washington (DC): National Park Service.

POWERS, R.P. 1984. Outliers and roads in the Chaco system, in D.G. Noble (ed.), New light on Chaco Canyon, Annual Bulletin of the School of American Research: 45--59. Santa Fe (NM).

RUSSELL, K.W. 1988. After Eden: the behavioral ecology of food production in the Near East and North Africa. Oxford: British Archaeological Reports. International series 391.

TAYLOR, C. 1979. Roads and tracks of Britain. London: Dent.

VAN LIERE, W.J. & J. LAUFFRAY. 1954. Nouvelle prospection archeologique dans la Haute Jezireh Syrienne, Les Annales Archeologiques de Syrie 4/5: 129--48.

WARBURTON, M. & D.K. GRAVES. 1992. Navajo Springs, Arizona: frontier outlier or autonomous great house, Journal of Field Archaeology 19: 51--69.

WILKINSON, T.J. 1989. Extensive sherd scatters and land use intensity, some recent results, Journal of Field Archaeology 16: 31--46. 1990a. Town and country in southeast Anatolia 1: Settlement and land use at Kurban Hoyuk and other sites in the Lower Karababa Basin. Chicago (IL): Oriental Institute. Oriental Institute Publications 109. 1990b. The development of settlement in the North Jazira between the 7th and 1st millennium BC, Iraq 52: 49--62.