The Archaeology of Suburbs and Sprawl

The definitions of what is a suburb and and what is sprawl require discussion, particularly in terms of varied modern versus archaeological use. Sociological and urban planning literature has reached no consensus on these definitions. Different sources define, describe, and classify modern suburbs along various dimensions, including location as well as functional, social, chronological, and organizational characteristics such as dominant mode of local transportation, racial or ethnic and class composition, visual appearance, degree of planning, types of ties to the city center, and chronological period or stage of development.⁴ At their root, however, these various definitions do typically agree on issues of location, newness, and density: a suburb is a habitation or use area at the periphery of a city’s core, founded later than the city core, and generally (but not always) with a lower density of occupation and activity than the city core and a higher density of occupation than the hinterland.⁵

In archaeological contexts, only some of the many dimensions often considered by sociologists and planners in the definition of suburbs are relevant and discernable. The modern association of suburbs with bedroom communities, automobiles, and mass transit obviously has no parallel in the preindustrial and ancient world.⁶ Although many define modern suburbs as primarily residential zones, many historical and ancient cities’ suburbs must have also been devoted to industry in order to keep noxious materials and odors away from the urban core and its residential zones.⁷ The political relationships between ancient suburbs and city cores, and the lines between former suburbs, active suburbs, and satellite communities of a particular historical period, can be difficult to establish in the absence of extensive written records. However, the base variables for defining suburbs (location, newness, and density) are archaeologically measurable and have indeed been observed among some ancient cities’ habitation zones.

The related term sprawl has also been inconsistently defined, and aspects of both its definition and supposed causes and effects on society remain much debated in scholarly contexts and the public arena. A review of social science definitions concluded that scholars differ in their relative emphases on sprawl’s density, continuity, concentration, clustering, centrality, nuclearity, mixed uses, and proximity.⁸ As for suburbs, some definitions of sprawl effectively assert that it could only be a modern phenomenon, for example definitions that emphasize the availability of transportation by car, lack of effective mass transit, and the prevalence of single-family homes. Other definitions are broader and explicitly encompass settlement forms in the historical and ancient past,⁹ for example the definition of sprawl as “low-density, scattered, urban development without systematic large-scale or regional public land-use planning.”¹⁰ Confusingly, sprawl is used in academic and public discourse to describe spatial patterns of human settlement and settlement processes that result in those patterns, as well as the causes and consequences of these patterns and processes.¹¹ When used to describe spatial forms or zones of settlement, sprawl often seems to refer to a type of suburb that exhibits a lack of planning in its organization or to urban extensions that spread (often discontinuously) beyond defined suburbs. In its use to describe settlement processes that result in low-density urban extensions, it is often used interchangeably with the terms urban growth and suburbanization.

This distinction between spatial forms and processes of settlement is significant for archaeology and the comparative historical study of urbanism, as the identification of similar settlement forms in the material record is much more straightforward than the inference of similarities in the dynamics that produced those settlement forms.¹² While it seems clear that some ancient cities had habitation and use areas that fulfill the location, newness, density, and lack-of-planning components underlying the majority of expansive modern definitions of suburb and sprawl as a spatial form, it remains unclear whether or not ancient suburbs and edge zones beyond them were formed in ways that can be considered analogous to modern urban sprawl as a growth process.¹³

The broad definitions laid out above may seem clear enough on the surface, but they also have crucial ambiguities. How far must a suburb be from the city center, and how far can it extend? How old or dense must a suburb become before it should be considered part of the urban core?¹⁴ Where is the line between suburbs and sprawl that may extend beyond it—how thin, diffuse, unplanned, or discontinuous must land use be for it to be considered sprawl rather than a defined suburb?¹⁵ There are no clear answers to these questions in modern contexts, let alone in archaeological ones.

For reasons of defense, many ancient cities were walled, and ancient suburbs are often defined simply as nearby extramural habitation and use areas. Because space is less constrained, extramural areas are often of lower density than crowded city centers. Given that undefended habitation was only possible during times of peace, the size and even location of extramural suburbs tend to fluctuate from period to period, producing horizontal stratigraphy.¹⁶ This definition and these generalizations may be irrelevant for unwalled ancient cities, and they are particularly inadequate for addressing instances of low-density urbanism¹⁷ that have in the last two decades become a focus of archaeological research in many regions of the world.¹⁸

Here, we leave those complexities and unresolved issues to the side, as our focus is on the early cities of southern Mesopotamia, the compact centers of which were typically walled.¹⁹ We thus apply the term suburb to relatively clearly defined habitation areas outside of the walled urban core, identified spatially and archaeologically via the intersection of variable topography, soil discoloration, architectural traces, and ceramic distribution (see below). We use the term sprawl in the first sense outlined above, to designate scattered, low-density, possibly unplanned, less clearly defined extensions of the ancient city beyond suburbs.

Even in such apparently simple cases of walled cities, the identification of suburbs and sprawl (as a spatial form) may be complicated by incremental city growth and city wall modification, which can result in former suburbs or peripheral zones being swallowed by the city core itself as it expands. Especially in cases of cities that persisted over centuries or millennia, as they did in Mesopotamia, archaeologists must be attentive to the fact that the intramural zones of a later period may be the extramural suburbs or sprawl of an earlier one. Likewise, the spatial contraction of a city may result in the establishment of suburbs in what were once core urban areas. In some archaeological cases, the distinction between suburbs considered part of the city and separate satellite settlements may be unclear without the availability of relevant textual data.

On top of these many conceptual issues, archaeologists and historians additionally face methodological problems in studying suburbs and sprawl. The identification and characterization of these areas on the ground in archaeological contexts requires spatially extensive survey²⁰ and excavation of residential and industrial structures. In the past, these data collection strategies were often at odds with an archaeological impulse to expand knowledge of monumental and elite constructions in the city center. The gathering of spatially extensive data to track urban growth and contraction is logistically difficult, time-consuming, and expensive when strata of periods of interest (particularly early periods near to the time of city foundation) are deeply buried under later habitation layers. At long-inhabited archaeological sites, the outlying areas of the city tend to shift through time, meaning that they often suffer more from erosion and other taphonomic processes than the deeply buried strata of the more continuously inhabited urban core. On the other hand, the potentially episodic nature of settlement outside the city core can make older habitation layers more accessible near the surface, without the complexities introduced by superimposed stratigraphy of subsequent phases and modification by later inhabitants.²¹ In the event that historical sources are available through excavation and other means, the biases of literacy, textual preservation, and material recovery often mean that the written record concerns activities, and relatively more affluent inhabitants, of the city center, with little information relevant for the reconstruction of suburbs and sprawl.

Ancient Mesopotamian Cities and Suburbs

The world’s earliest cities developed in southern Mesopotamia (modern Iraq) between the Tigris and Euphrates Rivers in the second half of the fourth to the third millennium BCE. Some neighboring areas in northern Mesopotamia (modern northern Iraq, northeastern Syria, and southeastern Turkey) experienced earlier proto-urban development in the fifth to fourth millennium BCE²² and adjacent lowland Khuzestan (modern Iran) saw roughly contemporaneous urban development in the fourth to third millennium BCE.²³ However, only in southern Mesopotamia did the early foundation of high-density cities result in an uninterrupted urban tradition that continued until the first millennium BCE and beyond.²⁴

Survey demonstrates that southern Mesopotamian cities were large. Uruk covered approximately 250 ha already by 3100 BCE and 400 ha by the early third millennium.²⁵ In the third-millennium BCE Early Dynastic period, many other cities, such as Lagash, Girsu, Umma, Shuruppak, and Kish swelled to 100–500 ha or more, depending on how their edges are defined.²⁶ Many of these size estimates remain uncertain because they rest on remote satellite imagery interpretation alone or ground observations collected decades ago, before the closure of Iraq to internationally collaborative research between about 1990 and 2012. Most 20th-century excavation data on early southern Mesopotamian cities focuses on civic and ceremonial architecture and other spaces inhabited or frequented by elite individuals. Some important exceptions include studies of the neighborhoods of the second and first millennia BCE at Ur, Nippur, and Mashkan-shapir.²⁷ Archaeological data on residential spaces is relatively limited, especially for the third millennium BCE and earlier,²⁸ and even less is known about suburbs, which have rarely been identified and even more rarely surveyed and excavated. In some cases, it is not clear whether or not existing city size estimates include lower-density, shifting suburbs or sprawl.

Work in northern Mesopotamia provides useful methodological and conceptual models for new investigations of southern Mesopotamian urbanism, including research programs geared toward understanding suburbs. Some of the only pre-1990s investigations of possible suburbs were carried out at the northern Iraqi sites of Tell Taya and Nuzi, where excavations revealed peripheral residential and industrial areas of the mid second millennium BCE.²⁹ Only at Tell Taya were the suburbs extensively mapped with survey and excavation, due to unique surface conditions that exposed the stone foundation of buildings’ walls.³⁰ Cities of the fifth to first millennium BCE in northern Mesopotamia often have a broadly similar form, with a high mound or citadel containing monumental structures surrounded by a residential so-called lower or outer town, typically many times larger than the citadel and often surrounded by a defensive wall.³¹ Since the 1990s, teams working at urban centers in northeastern Syria and southeastern Turkey have increasingly revealed the spatial and social structure of these outer towns using intensive, systematic surface survey, targeted excavation, and sometimes geophysics.³² These investigations have in some cases identified low-density or spatially discontinuous extramural habitation and use areas more distant from city cores that could be considered suburbs or sprawl. At Tell Brak, systematic survey showed that the city in the late fifth to early fourth millennium BCE had spatially discontinuous suburban areas that grew together over time,³³ and excavation revealed isolated industry, garbage, and burial areas.³⁴ At the massive 300 ha site of Tell Hamoukar, survey and excavation revealed dispersed, low-density residential and obsidian blade manufacturing areas, some perhaps used only seasonally in the late fifth to third millennium BCE.³⁵

The City of Ur

Ur is one of the southernmost, earliest, and longest-lived of the early Mesopotamian cities (fig. 1). Late 19th- and early 20th-century excavations at the site, especially the major campaigns of Sir Leonard Woolley in 1922–34, revealed an occupation history stretching over almost five thousand years, ca. 5000–300 BCE, from the Ubaid 3 period to the beginning of the time of Hellenistic and Seleucid hegemony in the area.³⁶ In an era when archaeologists tended to almost exclusively focus on monumental architecture in urban cores, Woolley’s excavations were remarkably broad. He did devote enormous effort to the excavation of Ur’s central religious precinct—containing the ziggurat (temple pyramid), temples, courtyards, storehouses, and administrative areas and residences of religious personnel—as well as the adjacent Royal Cemetery. However, his trenches also targeted areas removed from the core of public buildings, including residential neighborhoods of the second and first millennia BCE (Old Babylonian and Neo-Babylonian periods, especially areas AH, EM, and NH, with some other less well documented domestic contexts excavated in areas BC, CLW, EH, and SM), temples apparently located on the edges of the city (the Enki temple in the southeast), and other contexts. Woolley’s archaeological documentation of these residential and other nonmonumental contexts was admittedly highly uneven,³⁷ but his publications nonetheless provide a data set that remains unparalleled to this day. Thanks to Woolley’s broad-scale excavations as well as the analysis of cuneiform text archives recovered from them, Ur still provides one of the most detailed pictures that archaeologists have of daily life in third- and especially second-millennium BCE Mesopotamia.³⁸ Since 2015, new campaigns have expanded this picture even further, providing detailed modern excavation and laboratory data, as well as integrated archaeological and textual analysis for houses of the late third and early to mid second millennia BCE Ur III and Old Babylonian periods.³⁹

Located approximately 16 km southwest of modern Nasiriyah, Ur (modern Tell al-Muqayyar) today sits in the steppe-desert area, beyond the irrigated and cultivated lands surrounding the Euphrates River. Millennia ago, however, the city’s hinterland was radically different due to frequent avulsions (shifts) of the Euphrates in its lower reaches and significant changes in the sea and land levels; the gulf extended farther inland between ca. 8000 and 2000 BCE than it does today.⁴⁰ The ancient city lay along and between former branches of the Euphrates. The traces of these paleochannels appear clearly in archival aerial and satellite imagery and are still visible on undeveloped tracts of land (fig. 2).⁴¹ The dates of these channels are not directly known, though some of their ages can be delimited on the basis of smaller sites that lie along them, dated via ceramics collected in a regional survey in 1966.⁴² Additionally, throughout its habitation history, Ur likely was located near the coastline of the gulf and the marshes and lakes associated with the Tigris-Euphrates delta.⁴³

The exact timing of environmental changes and extent of the gulf and marshes remain unknown in relation to the settlement sequence at Ur. According to some landscape reconstructions and circumstantial evidence from texts, Ur in the late third and second millennia BCE may have been located upstream of where the Euphrates disappeared into the marshes, a location that made it the most downstream of the Mesopotamian cities and a major harbor for the region.⁴⁴ Recent geoarchaeological work at nearby Abu Tbeirah (15.5 km east of Ur) shows evidence of alternating marsh, floodplain, and riverine environments, though contamination by circulating carbon makes it difficult to absolutely date these changes. In the third millennium BCE, Abu Tbeirah is estimated to have been about 30 km from the gulf coastline, located within the marine-influenced Euphrates delta.⁴⁵

Ur today is located inside the Talil Air Base, and this has deeply affected the preservation of archaeological remains in both positive and negative ways. On one hand, the security offered by the base has meant that Ur’s main mound and some outlying areas have escaped the destructive looting that has dramatically impacted many other southern Mesopotamian sites.⁴⁶ On the other, the massive expansion of the base since the 1990s has resulted in the erasure of near-surface archaeological and paleoenvironmental features that had until recently remained preserved in Ur’s hinterland. More critically, military use of outlying areas beyond the main mound, especially areas to the northeast, have completely obscured archaeological features related to Ur’s suburbs and local hydrology.

Physically, Ur consists most visibly of a main oval-shaped mound measuring roughly 1,200 × 800 m and rising about 10–20 m above the level of the modern surface of the surrounding plain (figs. 3, 4). The edges of this mound appear clearly in satellite imagery and on the ground because they were bounded in the Ur III period (and perhaps other periods, see below) by a city wall, the remains of which have served in some places as a sort of retaining feature holding habitation strata of the mound in place. A steep-sided depression bisects the mound into a north and south half, and this depression may mark the former course of an intracity canal.⁴⁷ At the northeast corner and the west side of the mound, large depressions marked by light-colored sediment show the position of harbors contained within the city walls.

The excavations by Woolley revealed inscribed bricks that allow the dating of the foundation of the city wall to the Ur III period, and Woolley believed that by Kassite times the city was unwalled. It would make sense for other city walls of different periods to have existed at Ur. Prior to the Ur III period, Early Dynastic cities in southern Mesopotamia tend to be walled.⁴⁸ However, we currently have no evidence for earlier (or later) city walls at Ur (before or after the Ur III period) and in fact the information concerning the Ur III city wall from Woolley’s excavations is indeterminate in many ways.⁴⁹ Even if the city core was in fact unwalled in later times, the remains of the Ur III city wall were massive and must have endured as a visible marker of central urban space. Better information concerning the city wall or walls would likely significantly complicate our definition of the Ur city core, especially for pre–Ur III periods. For example, if the pre–Ur III city was unwalled, or the walled area was not extensive, perhaps confined to parts of the north half of the mound adjacent to the city’s major religious precinct, it is possible that other habitation areas such as those on the south half of the mound were at that point extramural suburbs. We do not address such possibilities here because our survey data are most suited for defining the extent of less deeply stratified outlying areas external to the Ur III period city wall. Nonetheless, it is important to acknowledge that our definition of the city core as coterminous with the Ur III walled area is a simplification of what was certainly a much more complicated situation through time.

Ur is both an example of and an exception to the overall lack of archaeological knowledge about southern Mesopotamian suburbs. Almost all archaeological attention at the site, from the 19th century to the present excavations, has focused on the main mound, the city core within the Ur III period walls. However, Woolley did observe at least nine extramural residential areas.⁵⁰ Most importantly, he topographically mapped mounds to the northeast and east of the main mound and devoted some effort to the investigation of an area locally termed Diqdiqqah, which he identified as a suburb of Ur. The area came to his attention because local workmen commuted across it as they traveled from their villages to the excavation trenches and turned in artifacts that they collected from the surface: terracotta plaques, cylinder seals, glazed frit objects, some jewelry, and stone-working tools.⁵¹ Woolley conducted some trial excavations here but abandoned the trenches after only a short time, believing that erosion and later cultivation had destroyed the stratigraphy and that the finds would not justify the expense of the work. Instead, he decided he could maximize the archaeological return for his funding by continuing to pay the workmen to turn over surface artifacts.⁵² After workmen reported the visibility of building foundations on the surface, Woolley later sent Max Mallowan to excavate one of the mounds at the northern edge, the exact location of which was not recorded. These excavations revealed the eroded remains of a Larsa-period building called the “Treasury” of Sin-Iddinam.⁵³ It was primarily through Woolley’s payments for surface finds that the Penn Museum (Philadelphia) and the British Museum (London) came to have some of the only artifact collections known to originate from an early Mesopotamian suburb.

Although the Diqdiqqah artifacts do not come from stratified contexts and do not have exact provenance, their characteristics and stylistic dating offer important clues to the chronology, environmental setting, and function of the suburb. Cylinder seals date mostly to the Ur III period, with a significant number of earlier seals and a few from the Isin-Larsa period and late second millennium BCE. The lack of inscriptions on the Ur III cylinder seals suggests that those living or working at Diqdiqqah in this period were of lower social status.⁵⁴ The large number of terracotta objects published by Woolley were dated by him primarily to the Ur III and Isin-Larsa periods (which in this case cannot be stylistically distinguished), with a few from the Kassite and Neo-Babylonian/Persian periods.⁵⁵ Further, Woolley noted that the majority of the surface pottery dated to the Isin-Larsa period, the same time period as the construction of the excavated Sin-Iddinam building.⁵⁶ The content of cuneiform texts and inscribed clay cones recovered from the surface suggested that three or four canals constructed by Ur-Nammu flowed in the area,⁵⁷ implying the existence of a weir serving a fan of branch canals.⁵⁸ Other loose inscriptions assumed to come from Diqdiqqah mention other canals and the sea.⁵⁹ Altogether, these artifacts led Woolley to conclude that Diqdiqqah was founded in the Ur III period, at the time of Ur-Nammu; saw particularly intense occupation in the early second millennium BCE and declining activity afterward; and had some level of continued occupation down to the Achaemenid period and the end of occupation at Ur.⁶⁰ The workmen recovered molds for the production of terracotta figurines, suggesting that the area was a craft quarter for the specialized manufacture of such objects, located where water and clay were available.⁶¹ Woolley also noted in unpublished reports that he believed Diqdiqqah had been an industrial area for the firing of bricks.⁶² Weights recovered among the surface assemblages suggest that the area participated in some sort of trade.⁶³ The hydrological situation of the suburb, the nature of its objects, and analysis of texts believed to refer to Diqdiqqah have suggested to some that it might have been a commercial district, karum in Akkadian, for the docking of boats and exchange of goods.⁶⁴ Texts indicate that the Ur harbor district was located outside the city walls, so it could not have been associated with the northeast and west harbors located in the city center itself. Diqdiqqah thus seems a likely location for it.⁶⁵

The observations of Woolley and others make clear that extramural settlement was not confined to Diqdiqqah and other suburb zones close to the main mound but instead that evidence of ancient activities sprawled far from the city center. In the 1850s, Taylor recorded the existence of mounds with slag and metal-working debris of unknown date approximately “one mile east” of the main mound.⁶⁶ Woolley mentions the presence of scattered ancient buildings to the northeast of Diqdiqqah (beyond the 1920s–30s Baghdad-Basra railway), to the east and southeast of an east mound stretching for 2 km, and throughout the roughly 6 km between Ur and Tell al-Ubaid to the west (see fig. 2). Unfortunately, all of these outlying areas of sprawl were not documented beyond a brief mention, and investigation of them today has been rendered impossible since they have been covered (and likely erased) by the Talil Air Base.

Woolley also considered the towns of Sakhariyah (multiple mounds about 5–8 km to the northwest of the Ur city center, sites 47, 50, and 53 in Wright 1981, covering roughly 7 ha total; see fig. 2) and Rejibah⁶⁷ (multiple mounds about 13 km to the southwest of the Ur city center, site 4–5 and site 93 in Wright 1981, covering roughly 23 ha) to be suburbs of Ur, but the distance of these sites from the city center makes it more likely that they were independent or satellite settlements.⁶⁸ In any case, they could not be considered suburbs in the sense that Diqdiqqah was, and no more so than the many other sites in the vicinity. Regional survey later showed the presence of about 45 archaeological sites within 10 km of the Ur city walls,⁶⁹ and satellite imagery analysis suggests there were at least another 20 more than that⁷⁰—most of these sites were unknown to Woolley in the 1920s–30s.

Beyond these general hypotheses, we know little about the suburbs of Ur and its extramural size, in part because archaeologists never subjected it to a systematic site and hinterland survey. As part of Robert Adams’ broad studies of the Mesopotamian plain, Henry Wright undertook a survey of the Ur-Eridu basin in 1966, published in 1981, that included visits to Ur’s main mound (site 10 in the 1981 catalogue) and Diqdiqqah (site 11–12), but the regional focus of this work and extensive, vehicle-based survey methodologies prevented him from carrying out a detailed investigation that would provide information on the growth and contraction of various sectors of urban settlement through time, comparable to data collected for Kish,⁷¹ Uruk,⁷² Lagash,⁷³ Mashkan-shapir,⁷⁴ and a few other southern sites investigated before the closure of Iraq to foreign archaeology between 1990 and 2012.

On the basis of his ceramic collections, Woolley’s previous excavations, and assumptions about the rate of site growth, Wright sketched a general and hypothetical settlement history of Ur and its suburb Diqdiqqah⁷⁵ (Table 1). Ur seems to have been founded in the Early Ubaid and grew by the Late/Terminal Ubaid to around 10 ha. Woolley only recovered Uruk-period material in an area used for ceramic manufacture,⁷⁶ but Wright supposed that Ur continued as a small town of 10 ha. By the Jemdet Nasr period, the area with Uruk ceramic manufacture was taken over by domestic buildings. Wright estimated that a similar expansion in all directions would have resulted in Ur’s growth to 15 ha, and then further growth to 21 ha in the Early Dynastic I period. In the Early Dynastic III period, archaeological and textual evidence of building activities as well as the famous Royal Cemetery suggest growth and prosperity at Ur under a dynasty of independent kings, and Wright surmised that the inner city might have reached close to its full extent of 50 ha. At the end of the Early Dynastic period, Ur’s independence briefly ended, and the region was under the control of other cities’ dynasties for a time, including that of Agade. During the Akkadian period of control, both Ur and Diqdiqqah (close to 5.9 ha in Wright’s approximation) were inhabited, but Wright found it difficult to estimate the settled area. At the very end of the third millennium BCE, Ur again had its independence and became the center of a regional territorial state or nascent empire (Ur III). It was during this period that the city wall still partially visible today was first constructed, enclosing roughly 50 ha. Diqdiqqah continued to be inhabited. Ur again lost its autonomy in the Late Larsa and Old Babylonian periods of the early to mid second millennium BCE, but the city and its suburb Diqdiqqah nonetheless became even more densely inhabited and occupied, expanding beyond the Ur III city wall. Wright estimated that the city covered a minimum of 60 ha but was not near to the 500 ha suggested by Woolley.⁷⁷ Following the Old Babylonian period, there was a large-scale regional abandonment. There is an absence of textual information for commercial activity at Ur itself ca. 1740–1400 BCE.⁷⁸ Wright believed that Kassite settlement after ca. 1400 BCE remained contracted inside the Ur III city wall. In the post-Kassite decline, after ca. 1200 BCE, Ur’s size continues to remain unknown, but Wright suggested that it might have decreased. By the first-millennium BCE Neo-Babylonian period, Wright postulated, Ur had become difficult to supply with water, and the city had low-density occupation perhaps covering 40 ha.

The edges of Ur and the maximal size of the site were defined by Wright as coterminous with the Ur III period walls bounding the main mound, encompassing approximately 50–60 ha. From a regional perspective, this size seems curiously small given Ur’s political importance and prominence as a major city, attested in the cuneiform textual record. During the third millennium BCE Early Dynastic period, Ur was a seat of important kings, and in the late third-millennium BCE Ur III period, it was the capital of a large territorial state or nascent empire that extended control over much of the southern Mesopotamian plain south of modern-day Baghdad. Other contemporary cities to the north such as Uruk, Lagash, Girsu, and Umma expanded in the Early Dynastic period to hundreds of hectares in size, and some of these cities seem to have maintained a large size into the second millennium BCE. Why, then, was Ur so small in periods when it was politically so important?

Declassified military intelligence imagery from 1959 and 1966 as well as UAV (drone) image mosaics captured in 2017 show soil discoloration extending hundreds of meters beyond the Ur main mound, suggesting the presence of large extramural settlement areas both in the direction of Diqdiqqah and elsewhere. Collectively, the main mound and areas of soil discoloration cover about 500 ha, and such a large area would likely have experienced shifting locations, forms, and intensities of use throughout and perhaps beyond Ur’s millennia of life as a city.⁷⁹ Emily Hammer hypothesized that these suburbs might significantly expand archaeological estimates of Ur’s size in the third to second millennium BCE and other periods, perhaps bringing Ur more into line with its peer cities. However, these settlement areas needed to be verified, mapped, and dated via systematic ground survey, and these hypotheses remained to be tested.

Hence, in 2016–17, Hammer began a site survey of Ur using remote sensing data and surface collections, with the goals of elucidating the settlement size and some aspects of city organization, as well as reconstructing the ancient hydrology in the city’s immediate hinterland. Specifically, the survey aimed to determine the extent of settlement areas off the main mound in various periods, to see if surface artifacts on the mound itself and the surrounding areas provide evidence for different types of activities or occupation in various periods, and to date relict watercourses and field systems around the site that had been mapped using aerial and satellite imagery. We report here on that survey and its results.

Ground Survey

The empirical definition of site edges and the identification of suburbs requires intensive distributional survey methodologies, like the nonsite survey methods pioneered in the Mediterranean that can trace gradations in settlement intensity or density over space.⁸⁰ The Ur ground survey, carried out in April 2017 and February–March 2019, consisted of collections of surface pottery and excavation of small test pits at 208 locations across parts of the Ur main mound and outlying areas (fig. 5a). The surface collections and test pits were spaced at intervals of 100 m in order to capture broad patterns across a large area. The team used submeter-accuracy GPS units with mobile GIS software to navigate to each predetermined survey point and to record survey information.⁸¹ A 10 × 10 m square was laid out with the survey point at the southwest corner of the square. All diagnostic sherds were collected within this 10 × 10 m square. Surveyors also noted the number of pieces of slag and the presence of other features (e.g., exposed walls, large amounts of decayed brick) within the square. Slag included fragments and nodules of vitrified clay and brick, highly variable in size and typically black or greenish in color. Many survey squares were carpeted in thousands of slag fragments (fig. 6), and in these cases, surveyors estimated the total number of fragments for the 10 × 10 m square by counting only the fragments in a 1 × 10 m area and multiplying by 10. Because the total number of sherds in many squares was also often too large to count and the number of sherds on the surface can be skewed by many factors, the team also dug a 50 × 50 × 50 cm test pit in the southwest corner of each square and collected separately all artifactual material from this small excavation (including undiagnostic sherds). The count and weight of all test-pit sherds and artifacts was noted and diagnostic sherds were kept. This general methodology is a modified version of that used in the site surveys of Tell Hamoukar⁸² and Tell Brak⁸³ in northern Mesopotamia and Lagash/Tell al-Hiba⁸⁴ in southern Mesopotamia.

The survey units cover the entirety of the fenced protected area enclosing Ur that has not experienced major modern surface disturbances. This fenced area includes about 310 ha of the roughly 500 ha total marked by soil discoloration in archival U-2 aerial and GAMBIT satellite photographs; the remaining such area was inaccessible because it is located outside the site’s fence, in the surrounding Talil Air Base, and the surface has been heavily obscured by developments on the base. On the main mound, the team only surveyed units located outside areas affected by previous excavations, earlier sediment dumps, and current or earlier heritage and touristic infrastructure. The survey area off the main mound, close to 135 ha, covers several topographically distinct features that had previously been identified with UAV photogrammetry.⁸⁵ These features include a substantial east mound (about 22 ha and 3–4 m high) located approximately 120 m from the main mound with substantial amounts of surface architecture; a chain of small mounds in the far northeast of the fenced area, within the area Woolley termed Diqdiqqah; a broad flat area with small, scattered, barely visible mounds to the north and northwest of the ziggurat with traces of a relict field system; and broad flat areas to the west and south of the main mound with traces of former watercourses and field systems (see fig. 4). A large area immediately to the north of the main mounds, separating the main mounds and Diqdiqqah, has been extensively disturbed by modern habitation and military activity in the last two decades,⁸⁶ especially the construction of military barracks to the northeast of the modern excavation houses. The number of bomb craters, modern building foundations, and remains of earth-working activities meant that it was impossible to survey in this area.

The summit of the east mound has been dug out for the construction of tank emplacements by the military. As the areas around these structures contained large amounts of chronologically diagnostic pottery, the team conducted a separate opportunistic collection of material to supplement the material collected as part of the normal 100 × 100 m survey grid. This opportunistic collection consisted of nine areal units (see fig. 5a).

Ceramic Analysis

Diagnostic sherds from surface collection units and test pits were brought to the Ur excavation house, washed, counted, and weighed. The team recorded ceramic fabric characteristics and vessel forms (open versus closed) for each sherd. All sherds were photographed, and a subset of sherds was selected for drawing. More detailed dating information for individual sherds was recorded later on the basis of photographs and drawings, in comparison with collections from excavations of other southern Mesopotamian sites, especially the excavated sequences from Uruk⁸⁷ and Nippur,⁸⁸ as well as other more recent studies of both previously and newly excavated material.⁸⁹ Angelo Di Michele has in recent years analyzed ceramics from the excavations of Tello, ancient Girsu,⁹⁰ and the Umma regional survey,⁹¹ so he was also able to make comparisons to unpublished, recent collections made via systematic, scientific fieldwork. Of 2,978 collected diagnostics, 1,271 were datable to one or more of 12 periods spanning from the late fifth millennium BCE to the 13th or 14th century CE (see fig. 5b).

Ceramic Dating

A full analysis of the dated sherds is in process; here we summarize the ceramic assemblage collected during the survey by identifying which diagnostic forms of each major period appeared most commonly (fig. 7). We also discuss which common ceramic forms extend over multiple periods.

Ubaid Period. Ceramics of the Ubaid period are characterized by black-on-buff pottery, mostly handmade or, to a lesser extent, made on a tournette or slow wheel.⁹² The survey assemblage includes various forms, including bowls with inwardly thickened rims with a high carinated profile (see fig. 7[1], 7[2]) and some examples with grooves on the lower internal side. Among the closed shapes are lugged jars (see fig. 7[3]). These sherds fit into a chronological horizon that extends from Ubaid 4 to Ubaid 5, as demonstrated by parallels with Abu Jarabie,⁹³ Tell Zurghul,⁹⁴ Tell Oueili,⁹⁵ and Uruk.⁹⁶

Fourth and Third Millennia BCE: Uruk, Early Dynastic, and Akkadian Periods. Mass-produced types predominate in the fourth- and third-millennia BCE ceramic assemblages. Uruk pottery from the survey mainly consists of fragments of the most common vessel of this period, the beveled-rim bowl (see fig. 7[4]).⁹⁷ The third millennium BCE is also represented mostly through mass-produced vessels, including solid-footed goblets (see fig. 7[5]) typical of the Early Dynastic I,⁹⁸ conical beakers (see fig. 7[6)] attested mostly in the Early Dynastic I–IIIA periods,⁹⁹ and conical bowls (see fig. 7[7]) attested, albeit with morphological variations, throughout the third millennium BCE.¹⁰⁰

From the Akkadian period onward (especially Late Akkadian) the ceramic assemblage from the survey becomes larger and more varied. In addition to conical bowls, the survey collected examples of a large bowl with an outwardly flaring thickened rim and a rounded carination (see fig. 7[8]).¹⁰¹ Some collected closed shapes, such as the low-necked jar with oval band rim (see fig. 7[9]), are attested during the entire Akkadian period.¹⁰² Other types spread over a chronological horizon encompassing the Late Akkadian period and the Ur III phase. This is the case of the low-necked jar with triangular rim (see fig. 7[10]), documented also in the ceramic assemblage of Lagash/Tell al-Hiba¹⁰³ throughout the Akkadian and Ur III phases.

Late Third and Second Millennia BCE: Ur III, Isin-Larsa, Old Babylonian, and Kassite Periods. The increase in pottery volume that began in the Late Akkadian period has its peak in the Ur III and Isin-Larsa periods. The most widely found shape of the Ur III and Isin-Larsa periods, the carinated bowl (see fig. 7[11]),¹⁰⁴ is abundant in the Ur survey. Another medium-sized bowl with an outside thickened rim and a high carination (see fig. 7[12]) is well documented in both phases¹⁰⁵ and is abundant in the survey. Other vessel shapes found in the survey are instead characteristic of a single period. For example, the jar with drooped, everted rim (see fig. 7[13]) is typical of the Ur III period.¹⁰⁶ The pottery of the Isin-Larsa period includes small bowls with high carination and a very small rim (see fig. 7[14]). This type is considered a variant of the main group of small carinated bowls that was widespread especially in the Middle and Late Isin-Larsa periods.¹⁰⁷ Closed shapes from the Isin-Larsa phase include the necked jar with ledge rim and with a well-marked carinated cylindrical body (see fig. 7[15]).¹⁰⁸

In the Old Babylonian period there is a marked decrease in the number of sherds, but the variety of represented shapes remains high. Goblets with sinuous walls and ring bases (see fig. 7[16]) are attested from the beginning of the period and represent a hallmark of the entire phase.¹⁰⁹ High-shouldered vessels with convex bases (see fig. 7[17]) also belong to the same chronological horizon and are well documented in the sites of southern Iraq.¹¹⁰ Some bases of these vessels were collected during the survey.

The number of diagnostics dating from the Kassite period is similar to that from the previous Old Babylonian period, but in the following Middle Babylonian period there is a significant decrease in the number of datable sherds. The Kassite period assemblage is dominated by solid or footed bases such as the shouldered goblets with a solid base, generally disc-shaped or slightly concave (fig. 7[18], 7[19]). This type of vessel is well attested throughout the period.¹¹¹ A few shapes are documented in both the Kassite and Middle Babylonian periods, such as small bowls with plain, rounded rims and more or less carinated walls (so-called wavy-sided) (see fig. 7[20]).¹¹² The pottery of the Middle Babylonian period is still poorly known in southern Iraq, so this period might be better represented in the Ur survey than we are able to identify at the moment.

First Millennium BCE. For the first millennium BCE, the collected, recognizable diagnostics date between the seventh and third centuries. In many cases these shapes persisted in the Neo-Babylonian, Achaemenid, and Seleucid periods. For example, the cup with an outside thickened ledge rim (see fig. 7[21]) is documented from the Neo-Babylonian up to the Seleucid phase.¹¹³ Carinated bowls, mostly glazed (see fig. 7[22]), are the most widespread type in southern Iraq from the fourth to the second century BCE.¹¹⁴ Jars have a series of thickened or bulb rims, sometimes with moulding on the neck (see fig. 7[23]). These narrow-necked jars were widespread from the Neo-Babylonian period to the Seleucid period and reached a peak in the Achaemenid phase.¹¹⁵

In the Ur survey, first-millennium BCE shapes dating before the Neo-Babylonian period (seventh century BCE) are not documented. The Neo-Babylonian period is mostly represented by closed forms, the Seleucid period is documented almost exclusively by long-lasting pottery shapes that cover a chronological span from the Achaemenid to the Seleucid period, and the shapes characterizing the Seleucid period are almost totally absent. The data collected seem to confirm an occupation during the first millennium that most likely started from the seventh century BCE, reached its peak in the Achaemenid period, and did not last beyond the third century BCE. Forms from subsequent periods (Parthian and Sassanian) are sporadically attested.

Ceramic Distribution

Maps of pottery density in the surface collection units (see fig. 5c) and test pits (see fig. 5d) as well as clay slag density in the surface collection units (see fig. 6) suggest that habitation continued for some distance off the main mound, especially to its east and northeast, covering much but not all of the 500 ha of soil discoloration visible in aerial and satellite imagery. In some areas, the edges of the soil discoloration seem to correspond best to the surface distribution of clay slag. The spatial distribution of pottery by periods (Table 2 provides the raw counts) shows the chronology of individual extramural habitation areas and allows us to stitch together a narrative of Ur’s expansion and contraction over millennia (figs. 8–10, and see Table 1).

Ubaid sherds (n = 26) are scarce and concentrated around the edges of the main mound and east mound, a pattern expected of material eroding from deeply buried layers in stratified areas. A light, broad scatter of Ubaid sherds covers about 35 ha to the northwest of the main mound, along the river paleochannel approaching the site from that direction (see fig. 8a).

Uruk sherds (n = 58) are slightly more numerous and found mostly at the edges of the main and east mounds as well as up to 300–400 m from their edges (see fig. 8b). The largest concentrations of Uruk sherds are located in the roughly 120 m area between these two mounds. The northern edge of Diqdiqqah yielded three scattered Uruk sherds.

Early Dynastic I and III sherds (n = 27) have been grouped for a distributional display, given the occurrence of many of the recovered forms in both subperiods. Only eight total identifiable sherds are specifically diagnostic of the Early Dynastic I subphase. Many Early Dynastic III forms continue into the early Akkadian period, so this period of major political change unfortunately cannot be recognized on the basis of ceramics alone. Early Dynastic III–Akkadian sherds (n = 46) are thus also included with the Early Dynastic group (n = 73). The Early Dynastic distribution suggests that this may have been the first time that significant settlement at Ur occurred away from the main and east mounds, marked by a significant scatter of sherds covering at least 3.5 ha on the northern edge of Diqdiqqah (see fig. 8c). In the vicinity of the main mound, Early Dynastic sherds are again mostly concentrated near the mound edges, as if they have eroded from deeper layers. Significant scatters of Early Dynastic pottery also occur in topographically flat areas to the west of the main mound, covering about 17 ha, and across the center of the east mound.

Late Akkadian sherd scatters (n = 76) are concentrated on the northern edge of Diqdiqqah (about 14 ha), the northwestern edge of the main mound, and across the length and width of the east mound (see fig. 8d).

Ur III period sherds (n = 295) show the first broad distribution across much of the surveyed area, including the interior of the main mound, the east mound, the space between these mounds, the north edge of Diqdiqqah (covering about 15 ha), and flat areas to the northwest, west, and south of the main mound (see fig. 9a).

This broad distribution continues in the Isin-Larsa period (n = 293), although the extent of the scatter on the east mound is considerably smaller and more restricted to the central area in comparison to the broader Ur III period scatter (see fig. 9b).

Old Babylonian sherds (n = 107) are mostly concentrated in the south portion of the main mound and its edges and in the central part of the east mound (see fig. 9c). The sherd scatter at Diqdiqqah is greatly reduced in quantity and extent in comparison to the Ur III and Isin-Larsa periods.

Kassite-period sherds (n = 81) are numerous in the central part of the east mound and across the southern extent and edges of the main mound (see fig. 9d). Sherd scatters of this period are almost totally absent at Diqdiqqah. Only 17 Middle Babylonian sherds of the later second millennium BCE were identified.

First-millennium BCE sherds (of the Achaemenid, Seleucid, and Parthian periods, n = 238) are again broadly spread across the surveyed area, including the edges of the main mound, the central part of the east mound, Diqdiqqah (about 6.5 ha), and flat areas covering roughly 42 ha to the west and northwest of the main mound (see fig. 10a). Notably, the sherd scatter to the west and northwest of the main mound is considerably denser and more extensive than in any other period. The dense scatter covers the entirety of the area in which relict canals and field systems are visible in satellite and aerial imagery. As would be expected given what we know of Ur’s abandonment around the end of the first millennium BCE, sherds of later time periods are extremely scarce. The survey identified only five sherds of the Middle–Late Islamic period (see fig. 10b).

Many factors other than the intensity, chronological length, and spatial extent of habitation of a particular period can shape surface ceramic distributions. These factors include the presence or absence of later habitation and how deeply buried strata of a particular period become, the chronological length of a ceramically defined archaeological period, the distinctiveness and length of use of that periods’ ceramics, erosion and other taphonomic processes, and human disturbance of the site for excavation, tourism, and other purposes. At Ur the huge sediment dumps from old excavations (as well as erosion from these dumps and earth-working of the dumps to create touristic infrastructure) are major factors shaping surface sherd distribution. Thus, the survey avoided collection near disturbed areas (as discussed above, see under “Methods”). Additionally, we primarily use the results to inform our understanding of the settlement history of the extramural areas that tend to be comparatively less impacted by many of these processes.

East Mound

The east mound appears in Woolley’s topographic map, but he did not excavate there, and other reports on the site have frequently not even noted the presence of this mound. Survey thus has much to contribute to our knowledge of this very underappreciated but likely important part of the site (fig. 11). Finds were collected from 25 surface collection units and test pits across the east mound and surrounding lower-lying areas, as well as from nine opportunistically placed areal collection units on the summit of the mound, the positions of which are marked in the figures. The mound itself, close to 12 ha at its center with northern and southern extensions bringing its area to about 22 ha, rises a maximum of 3–4 m above the average surrounding ground surface at its center and 1–2 m above the surrounding terrain in its northern and southern extensions. The mound’s surface is covered with baked brick alignments that likely indicate the outlines of Old Babylonian houses, streets, and drains. These baked brick alignments were previously mapped using drone photography in 2017 (see fig. 3b).¹¹⁶ Further aerial mapping in 2019 and thus far unpublished magnetometry data¹¹⁷ indicate the presence of a large building, at least 45 m across, located at the summit of the mound. The magnetometry data additionally indicate that dense architecture was present across at least a 130 m wide area (west-east) at the center of the mound.

The distribution of Ubaid and Uruk sherds suggests that the east mound of Ur could have been established at or near the same time as the main mound itself. The distribution of sherds across the east mound in later periods suggests that east mound settlement may have rapidly expanded in the Early Dynastic–Akkadian and especially the Ur III periods. Settlement perhaps contracted in Kassite times. First-millennium BCE activity may have slightly expanded around the east mound.

Northeast: Diqdiqqah

The collections near Diqdiqqah, unfortunately mostly confined to the northern edge, nonetheless help clarify the occupational history of this extensive suburb. Finds were collected from 34 surface collection units and test pits across a roughly 25 ha area at the far northeast corner of the site’s fenced area. Extremely dense scatters of highly abraded sherds characterize this flat area (fig. 12). Only about half of the collected units fall within the soil discoloration area visible in U-2 photographs and GAMBIT imagery (fig. 13). Much of the area with clear soil discoloration marking Diqdiqqah was not accessible (outside the fence) or too disturbed by recent activity for survey. For these reasons, the survey data do not show the full extent of ceramic scatter and make it impossible to empirically trace the size of Diqdiqqah through time. The extent of ceramic scatter visible in the distributional maps only provides minimum estimates.

The survey suggests that occupation here may have begun considerably earlier than previously believed, by the Early Dynastic period (a few Uruk sherds might hint at even earlier activity), with a major expansion in the Late Akkadian period. The data support Woolley and Wright’s conclusions that Diqdiqqah was particularly important in the Ur III and Isin-Larsa periods, with a decline or spatial contraction of activity afterward, from the mid to late second millennium BCE. However, the broad distribution of first-millennium BCE ceramics suggests the return of activity to this area during the final centuries before Ur’s abandonment.

The collected material and other observations not yet reported in the accompanying maps show that the Diqdiqqah area is spatially divided between kiln areas with large amounts of brick and clay slag and other activity areas with almost no slag, but very high densities of pottery, approaching and even surpassing the densities of surface pottery on the main mound (see fig. 13b, c).

Aerial and satellite imagery, especially U-2 imagery, clarify the topographical structure of Diqdiqqah before the area experienced major surface disturbance (see fig. 13a). The northeastern suburb appears to have two or possibly three separate areas: a triangular area of about 54 ha with multiple mounds extending from the north edge of the east mound to the vicinity of the old Baghdad-Basra railway, a separate mound of about 9 ha lying to the east of the railway, and a possible southern extension of this mound of about 4 ha. The largest, triangular area has at its center a line of low mounds stretching south-north. Some these mounds are distinct and others are interconnected, but the U-2 image shows at least 14 peaks or local high points. The linearity of these mounds’ positions and their association with a concentration of clay slag continuing toward the north (see fig. 13c) suggests that they might represent discrete areas of industrial activity (perhaps workshops or dumps?) along a road or watercourse. The features in the U-2 image suggest that Diqdiqqah’s spatial connection to the urban core of Ur may have been via the east mound. If the soil discoloration in the imagery is indeed indicative of occupied area, there appears to have been a gap between the Diqdiqqah triangle and the main mound, and in any event, the northeastern edge of the main mound was occupied by one of Ur’s city harbors.

Relict watercourses visible in the imagery in the vicinity of Diqdiqqah might be considered possible candidates for the canals mentioned in the texts (see fig. 2). Two sinuous watercourses run on the north side of the soil discoloration, and these watercourses head directly toward Wright’s sites 16–20, Tells Sughariyya and Ba’arura, located 3–5 km to the east of the edge of Diqdiqqah (5–7 km from the Ur city center). These nearby sites appear to have been inhabited in the third to late second millennium BCE, from the Early Dynastic through the post-Kassite periods, though not all at once. The chronology thus fits with the idea that these watercourses were active during the times we now know Diqdiqqah to have been inhabited.

Northwest, West, and South Areas: Watercourses, Field Systems, and Low Mounds

The areas to the northwest, west, and south of the main mounds are much flatter than the east and northeast regions, characterized only by very small, low individual mounds. These mounds (roughly 0.25–0.50 m high) have been mapped both on the ground and with UAV photogrammetry. Most occur immediately adjacent to the traces of former watercourses, and most are aligned to the northwest of the ziggurat. The general lack of topographic expression and the presence of paleochannels and relict field systems in these broad areas could suggest that they only saw episodic or ephemeral habitation, or alternatively could suggest that these were areas used for agriculture, herding, or other spatially extensive activities rather than habitation. Finds were collected from 79 surface collection units and test pits across the west edge of the main mound and low, flat areas to the west and northwest, covering about 36 ha. The pottery from these units is frequently small and often abraded. Another 36 units were collected to the south of the main and east mounds, covering about 22 ha.

The Ubaid and Uruk scatters in these areas are light, and the first potential evidence for significant activity in these areas occurs with a roughly 17 ha concentration of Early Dynastic pottery to the west of the main mound. For the subsequent Late Akkadian, Ur III, and Isin-Larsa periods, scatters of significant spatial extent and density continue to occur in this western area of 8, 26, and 20 ha, respectively. The Ur III and Isin-Larsa scatters extend farther to the north, along the major northwestern paleochannel flowing toward the main mound (fig. 14). Old Babylonian–period scatters suggest a spatial shift in extramural activity to the south and southeast, perhaps indicating that habitation grew between the south portion of the main mound and the southern extension of the east mound. Kassite and first-millennium BCE scatters again occur to the west of the main mound. The pottery scatter of the first millennium BCE is significantly broader and denser across the west and northwest areas than that of any other time period, covering close to 42 ha.

Because of the extensive and overlapping pottery scatters of many periods, it is impossible at this point to hypothesize the dates during which the visible paleochannels and relict field systems to the northwest, west, and south were active. Excavation and absolute dating will be necessary to better constrain the chronology of these features.