Khmer settlement patterns

Researchers have studied Angkorian settlement and urbanism for nearly a century (e.g. Coe Reference Coe1957; Groslier Reference Groslier1974; Pottier Reference Pottier2000, Reference Pottier and Haendel2012; Gaucher Reference Gaucher2004; Evans et al. Reference Evans, Pottier, Fletcher, Hensley, Tapley, Milne and Barbetti2007), and growing evidence from the pre-Angkorian period indicates that Khmer socio-political organisation extends back 200 years before the founding of Angkor (e.g. Vickery Reference Vickery1998). Figure 1 locates Angkor Wat, Greater Angkor and the LOMAP survey area. Few researchers have documented Angkorian residential space (e.g. Gaucher Reference Gaucher, Karlström and Källén2003; Bâty Reference Bâty2005); none have studied households as an analytical unit. As archaeologists define ‘household’ differently (Pluckhahn Reference Pluckhahn2010: 338–39), we focus on co-residential groups that used occupation surfaces, features and artefact assemblages associated with dwellings that stood on some or most mounds visible throughout the outer enclosure (e.g. Nash Reference Nash2009: 224).

Figure 1. Map of Greater Angkor, including the sites mentioned: inset top left, regional view; inset bottom left, detail of the central urban area; data courtesy of NASA-SRTM, JICA, Damian Evans and Christophe Pottier.

We use the term ‘settlement configuration’ to describe clusters of archaeological features that recur across the landscape and circumscribe larger social arrangements (see also Evans Reference Evans2007: 154–58). Settlement configurations are surface phenomena, measurable using ground-survey techniques and airborne and spaceborne remote-sensing technologies. We readily acknowledge that our conceptual framework overlooks both short-term and long-term spatial activities described previously in the literature (e.g. Johnstone Reference Johnstone2004; Stark Reference Stark2006b).

At its height, the Angkorian state comprised a complex network of settlements, craft-production centres, transportation linkages and resource areas whose landscape stretched into modern-day Vietnam, Laos and Thailand (e.g. Groslier Reference Groslier1973; Pottier Reference Pottier1999; Stark Reference Stark2006b; Fletcher Reference Fletcher2009; Hendrickson Reference Hendrickson2010). From the ninth century AD onwards, most Khmer urban space was characterised by a dispersed, low-density spatial structure that created the sprawling urban complex of Greater Angkor (e.g. Fletcher Reference Fletcher and Smith2012; Evans et al. Reference Evans, Fletcher, Pottier, Chevance, Soutif, Boun Suy, Sokrithy, Darith, Tina, Samnang, Cromarty, De Greef, Hanus, Bâty, Kuszinger, Shimoda and Boornazian2013; Hawken Reference Hawken2013).

Traditional Khmer settlement was agrarian and rural in nature (Delvert Reference Delvert1961); archaeological work in southern Cambodia indicates it may have substantial time-depth (Figure 2). Four field seasons of survey by LOMAP covered 35km² and identified three modal settlement configurations for the pre-Angkorian period. The first involved small clusters of mounds with associated ponds and moated mounds—or “moat-and-mound temple-centred” configurations (Evans Reference Evans2007: 24–26). These could be hamlets and conform to the phum (village-level) administrative, settlement and work unit found today across Cambodia (Delvert Reference Delvert1961). LOMAP also mapped large sites with multiple moated mounds or mound-and-pond complexes (which we interpreted as phum or ‘villages’). A third settlement form involved a series of dispersed moated mounds. Figure 3 illustrates examples of pre-Angkorian settlement units from the Mekong delta and Angkorian settlement units from the Tonle Sap area. All three modal forms include mounds and ponds (the modern Khmer equivalent for pond is trapeang), around which several households lived to access fresh water during the prolonged annual dry season, and which characterise rural Khmer settlement today.

Figure 2. Modal Khmer residential units: top left, a typical moated local temple site; top right, moated local temple site near the high-water mark of the Tonle Sap lake at Angkor, surrounded by occupation mounds; bottom, digital terrain model of a local temple surrounded by occupation mounds and ponds in the Roluos group at Angkor; derived from LiDAR data with vegetation and other non-ground elements filtered from the scene, and with brighter areas indicating higher elevation; LiDAR courtesy of KALC.

Figure 3. Khmer settlement patterns: top, features mapped in Preah Phkoam area (southern Cambodia, from Stark Reference Stark2006b); bottom, features mapped in the Greater Angkor region by Evans (Reference Evans2007; background data in both images is 1:25 000 scale aerial photography acquired in 1992–1993 by FINNMAP.

These early to mid-first millennium AD LOMAP settlement patterns contain features that pre-Angkorian inscriptions describe, such as residential areas, water tanks, paths/roads/causeways, groves and plantations, and moated temples (or prasat) (Jacob Reference Jacob, Smith and Watson1979: 413). Epigraphic data indicate some variability in dispersed settlement form; for example, some villages were attached to particular rulers or temples (Sahai Reference Sahai1977: 47), while others may have been more autonomous. Archaeological research in north-west Cambodia suggests settlement pattern continuity into the Angkorian period (Evans Reference Evans2007: 184).

Angkorian period urbanisation produced new settlement configurations: the orthogonal grid, and perhaps also linear settlement along rivers, canals and roads (Pottier Reference Pottier and Haendel2012). Some relatively early temples in the Angkor region (e.g. Prei Monti, Preah Ko and Bakong) are also surrounded by multiple outer enclosures, but these may have primarily served non-habitation functions. For example, the eighth- or ninth-century Bakong temple functioned as a largely monastic (not residential) complex during its primary occupation, with residential mound clusters distributed for some distance outside its moats in a low-density, dispersed pattern. Other key ninth- to mid-eleventh-century temples that may have had associated residential areas lacked ‘city’ enclosures (e.g. Bakheng, Ta Keo and Baphuon) (Pottier Reference Pottier2000, Reference Fletcher and Smith2012: 21).

Twelfth-century Angkorian kings constructed a series of temple enclosures on large-scale orthogonal grids (e.g. Gaucher Reference Gaucher, Karlström and Källén2003, Reference Gaucher2004; Pottier Reference Pottier and Haendel2012; Evans et al. Reference Evans, Fletcher, Pottier, Chevance, Soutif, Boun Suy, Sokrithy, Darith, Tina, Samnang, Cromarty, De Greef, Hanus, Bâty, Kuszinger, Shimoda and Boornazian2013). Within its moated and walled enclosure, the Angkor Wat orthogonal grid pattern divides the space into blocks, each of which are further systematically divided into mounds and depressions that functioned as ponds. Twentieth-century surface modifications to the enclosure have obscured parts of the patterning, but we estimate that the Angkor Wat enclosure contained approximately 283 mounds and 250–300 ponds. Field-based investigations during 2010 and 2013 examined the nature and time-depth of this Khmer grid-structured settlement configuration from the twelfth century onwards.

The Angkor Wat 2010 and 2013 excavations

Excavations were undertaken at Angkor Wat from June to August 2010, and from June to July 2013 in collaboration with the APSARA (Autorité pour la Protection du Site et l'Aménagement de la Région d'Angkor) Authority. The 2010 work involved ground-verifying GPS prospection patterns in the enclosure's western section (Sonnemann et al. Reference Sonnemann, O'Reilly, Rachna, Fletcher and Pottier2015: 1420). Work also focused on the less disturbed eastern enclosure and areas immediately adjacent to Angkor Wat's eastern moat, presented here as grid maps superimposed on a LiDAR digital terrain model (Figure 4). The primary research objective was to characterise the nature and dating of residential use through topographic and sketch mapping, systematic coring and stratigraphic excavations.

Figure 4. LiDAR image of formal grid system across Angkor Wat enclosure: top, merged digital terrain model and hillshade model; bottom, digital terrain model with overlay of the grid reference system devised by Heng Piphal; brighter areas indicate higher elevation; LiDAR courtesy of KALC.

Fieldwork

Mapping in 2010 began in the north-east quadrant with a 40 × 40m grid, which revealed a system of mounds and depressions. Few surface artefacts were visible, with the exception of a few laterite masonry fragments near unit 7. Subsequent acquisition, processing and use of LiDAR data for GAP 2013 fieldwork in the Angkor Wat enclosure confirmed this 2010 field-based mapping. The 2010 coring transects using east–west and north–south profiles of the eastern enclosure helped identify intact archaeological deposits and build an area-wide stratigraphic sequence. The 2013 coring of depressions was ground-truthing for evidence of long-term standing water that would support the classification of these depressions as ponds. Every depression visible in the LiDAR and DTM projection of the south-east quadrant was sampled.

Fieldwork also included stratigraphic excavations of 25 trenches (6 in 2010; 19 in 2013) on mounds, along mound slopes, in depressions and adjacent to walls; 11 trenches sampled the tops of mounds. In 2013, three trenches were also excavated east of the outer wall and moat, where LiDAR imagery revealed a rectangular enclosure. Figures 5 and 6 identify trench locations in the north-east and south-east quadrants within the walled enclosure, and Figure 4 identifies trench locations in this outer eastern enclosure.

Figure 5. Comparison of maps of the north-eastern quadrant of the Angkor Wat enclosure, from before and after the 2012 LiDAR mission (Evans et al. Reference Evans, Fletcher, Pottier, Chevance, Soutif, Boun Suy, Sokrithy, Darith, Tina, Samnang, Cromarty, De Greef, Hanus, Bâty, Kuszinger, Shimoda and Boornazian2013). Top, ground survey (courtesy of Shawn Fehrenbach), superimposed onto image (from Sonnemann et al. Reference Sonnemann, Sauerbier, Remondino, Schlotter, Campana and Forte2006). Bottom, merged digital terrain model and hillshade model derived from the 2012 LiDAR data with vegetation and other non-ground elements filtered from the scene.

Figure 6. Locations for all excavation trenches in eastern quadrant excavated in 2010 and 2013; grid lines identified through LiDAR are indicated in red (image courtesy of Heng Piphal).

Stratigraphy

Figures 7 and 8 illustrate the four layers that characterised most trenches across the eastern enclosure. Layer 1, an organic and loamy topsoil, whose basal section contained some cultural materials; layer 2, a cultural matrix that contained ceramics (earthenware, stoneware and high-fired Chinese tradewares), some postholes, flat-lying stones or pot breaks in one or two sub-layers that represent surfaces; layer 3, a thick cultural lens of sandy clay with manganese inclusions; and layer 4, the natural substrate, which is even sandier clay. Trenches closest to the temple contained sandstone-chip construction-related debris as part of layer 2; trenches abutting the enclosure wall had the lowest artefact densities in layer 2. Layer 3 appears to represent the initial construction of the mound-pond grid structure and is associated with the temple's construction. Layer 2 is associated with the reign of Suryavarman II and use of his Vishnu temple. No pre-Angkorian signature was identified through the extensive coring and trench excavations in the eastern temple enclosure.

Figure 7. General stratigraphic sequence in eastern portion of Angkor Wat enclosure, illustrating layer 2 as occupational layer; layer 3 as possible construction layer (note trench 17, north wall), and pond feature in trench 24b, south wall (image courtesy of Chhay Rachna).

Figure 8. Photograph of trench 17 north wall.

Accordingly, layers 2 and 3 are the focus of this discussion. Layer 2 in many trenches contained charcoal flecks, ceramics and recycled architectural stone, and possibly recycled architectural ceramics. Trench 3 yielded an earthenware stove (or brazier (Cremin Reference Cremin2009: 81)) fragment. Most earthenware sherds (except a single stove fragment) derived from utilitarian vessels such as cooking pots and water jars; some sherds that showed evidence of wear or sooting were probably cooking pots. Stoneware sherds derived from jars, baluster and pedestal vases, are associated with Angkorian-period occupation and are also illustrated in the bas-reliefs of the twelfth-century Bayon temple. Khmer glazed stonewares were recovered across the Angkorian landscape, and include storage jars, boxes, bowls and bottles (Guy Reference Guy1997: 55–58; Cort Reference Cort, Cort, Farhad and Gunter2000: 98–108; Chhay et al. Reference Cremin2009, Reference Rachna, Piphal, Visoth, Klokke and Degroot2013; Ea Reference Ea2010).

Small fragments of high-fired Chinese ‘tradeware’ ceramics were also recovered during the excavations. Chinese tradewares recovered from archaeological sites across Greater Angkor date primarily from the Northern Song (AD 960–1279) through to the Ming (AD 1368–1644) dynasties (Ea Reference Ea, Pei-Kai, Li and Wan2005; Cremin Reference Cremin2006). Some particular Chinese production centres, such as the Guangdong kilns, manufacture goods that probably served as diplomatic trade products (Wong Reference Wong1979), and have been recovered in elite and non-elite contexts throughout the Angkorian realm (Wong Reference Wong2010). Lower quality Chinese bowls, boxes and globular vases are also associated with residential sites in the Angkor region (Groslier Reference Groslier1981: 230–31).

Age of the habitation deposits

Khmer people used the Angkor Wat enclosure during and after the Angkorian period. Angkor, including Angkor Wat, housed Buddhist sanctuaries that attracted pilgrims from various parts of Asia during the post-Angkorian or ‘middle Khmer’ period (Loureiro Reference Loureiro2005: 218; Sonnemann Reference Sonnemann2011: 44). One of several sixteenth- to eighteenth-century inscriptions found on walls and pillars at Angkor Wat, dating to 1579, calls the temple Visnuloka and records the restoration of the temple's walls and roof (Lewitz Reference Lewitz1972: 109). The identification of a seventeenth-century Japanese map of the Angkor Wat temple, and other post-fourteenth-century remains, underscores Angkor Wat's centrality as a religious centre for centuries after the Angkorian capital moved south-east (Thompson Reference Thompson, Jessup and Zephir1997: 28–32; Sonnemann Reference Sonnemann2011: 137–38).

The 2010 and 2013 excavations produced both Angkorian and post-Angkorian materials in most trenches; the trenches along the enclosure's wall contained more post-Angkorian material. Two trends are clear in Figure 9, which presents radiocarbon dates from the 2010 excavations to characterise the occupation history of the Angkor Wat enclosure. First, locations near the temple and in the central north-east quadrant (i.e. unit 7) were used during the twelfth-century AD construction and initial use of the Angkor Wat temple. This pattern, coupled with LiDAR evidence for contemporaneous use of areas beyond the enclosure's eastern moat (Fletcher et al. Reference Fletcher, Evans, Pottier and Rachna2015: 1393), defines the broad area that included Angkor Wat and its environs. Secondly, post-Angkorian Khmers also lived in the enclosure, as evidenced by radiometric dates (trenches 5, 6 & 8) and the recovery of Ming period tradewares. The fifteenth-century political transformation of the Angkorian state did not therefore entail an abandonment of the Angkor Wat temple area, as often assumed (see also Brotherson Reference Brotherson2015).

Figure 9. Probability distributions of (2σ) calibrated radiocarbon samples from GAP excavations at Angkor Wat (courtesy of David Brotherson), with distinct phases highlighted in pale orange; left, Angkor period construction and use of the enclosure and surrounding area; right, post-Angkorian use of the enclosure from the fifteenth to eighteenth centuries).