Monash University Publishing | Contacts Page
Monash University Publishing: Advancing knowledge

Old Myths and New Approaches: Interpreting Ancient Religious Sites in Southeast Asia

Chapter 6

Connecting the dots

Investigating transportation between the temple complexes of the medieval Khmer (9th–14th centuries CE)

Mitch Hendrickson

Assistant Professor, Department of Anthropology, University of Illinois at Chicago

There is no real history if one does not account for the way a group has organised its natural space, just as a geographer must follow the birth of a civilization step by step in order to understand how it came about.
(Groslier 1986[1973]:31)

[Transportation] is an agency by which every part of society is brought into relation with every other, and interdependence, specialization, in a word, organization, made possible…
(Cooley 1974:28)

The Khmer temples of modern day Cambodia, Laos, and Thailand have been the religious and iconic foci of the people who lived around these places for over a millennium. Not surprisingly, academic interest in the region has similarly been drawn to the vast array of architecture, art and epigraphy found at these sites. A consequence of this is that Khmer temples have been largely studied in isolation from each other and their regional geographic setting. B-P Groslier (1986[1973]) argued that any rational history of the Khmer and their temples must take into account the physical landscape, moving away from the local to consider patterns at regional and supra-regional scales. When viewed in this manner, the arrangement of these temples across the landscape provides crucial information about the nature, expansion and changes of the Khmer empire.

The approach used in this investigation is an examination of spatial relationships between temples through a theory of transportation and communication. The need for movement of goods, people and ideas is linked to every aspect of society, and, as with the historic and modern nation states, binds society (or societies) together (Cooley 1974:28). Thus, it is logically informative to examine how the Khmer temples were integrated into an ‘empire’ using principles of transportation as the means of connection. By using geographic information systems (GIS), preliminary models of how and where contemporaneous Khmer temples ‘communicated’ with each other and how the site selection is intricately linked to the need for transportation are examined.

The role of transportation theory

Transportation is one of the most pervasive components in society and can influence or be influenced by any combination of economic, religious, political, military and social factors. The study of transportation developed within transport geography and is specifically concerned with the movement of people, freight and information across the landscape. A transport geographer views the organisation and relationship of three central components—nodes (locations), networks (linkages between nodes) and demand (human need)—normally in relation to economic considerations related to infrastructure, logistics and specialisation (Rodrigue, Comtois & Slack 2009). For the purposes of this study, the role of transportation theory is limited to identifying actual and potential linkages (land or water routes) between a select group of nodes (the Khmer temples). If, as assumed, the Khmer temples are part of a greater cultural landscape under politico-religious control, an examination of connections at the regional and supra-regional scales will provide insight into the internal dynamics of the Khmer empire and its geographic history.

Evidence of Khmer transportation

Before analysing the transportation connectivity between temple nodes, it is worthwhile to outline the amount of land and river transportation information available from the archaeological and historical records. The Angkorians left significant evidence of their terrestrial transportation system, though it was not as substantial as the Roman road network. In the late 19th century, earthen road embankments and associated stone bridges were rediscovered during French reconnaissance missions (Aymonier 1900–01; Lunet de Lajonquière 1902, 1911). Artistic evidence of Angkorian mobility is also found on the bas-reliefs of the Bayon, Angkor Wat, and Banteay Chhmar temples with repeated depictions of different types of transport employed at that time (for example, elephants, chariots, horses and ox carts).

Perhaps the most important addition to our knowledge of the history and composition of land transportation came with the publication of Georges Càdès’ (1941) translation of the Preah Khan stela. Dated to the reign of Jayavarman VII (late 12th century CE), it describes three roads leading from Angkor to provincial centres and the construction of 121 rest-houses along these roadways. This information led to the rediscovery of many stone buildings, now referred to as rest-houses, placed at intervals of 12 to 15 kilometres along the roadways (Ittaratana 1998). More recently, a similar pattern was identified in the spacing of rectangular water tanks (trapeang) at four to five kilometre intervals along most of the Angkorian roads (Hendrickson 2004). This combined information points to an elaborate road network with a centrally-planned infrastructure to support the regular movement of people across the region.

Though it seems less visible in the archaeological record, water transportation played an important role in Khmer mobility. Rivers provide obvious natural routes of communication that would have been exploited long before the Angkorian period. Unlike roads and their associated infrastructure, which are highly visible archaeologically, rivers require little capital investment. It is also difficult to determine when, where, how and why they were used. The strongest evidence supporting the use of rivers as key transport routes is the placement of Khmer temples along major waterways (Groslier 1973:117; Im 1998:57–72). Hall (1985:173) argued that this pattern reflects the economic importance of rivers, in particular the Sen River in central Cambodia, as conduits for trade. Physical remains of water transport are currently restricted to the boats depicted in the battle scenes from temple bas-reliefs and landing platforms associated with the temple reservoirs, such as those at Srah Srang and Phimai. Interestingly, Paris’ (1941b) study of these images concluded that the design of the boats is only suitable for riverine transport. While seemingly obvious, river transport must be considered to have had a significant role in day-to-day movement during the Angkorian period and must be thoroughly studied to understand past Angkorian societies. What is needed currently is to examine how the road and water routes were integrated into a regional transport network.

Transport theory in a Khmer context

While there is ample evidence of terrestrial transportation infrastructure to which transportation theory is readily applicable, the first stage of any such analysis is to understand the synergy between roads, rivers and temples. The primary aim of this chapter is to use transport theory to model potential locations of communication routes between the Angkorian temples. If our assumptions are correct these temples were integrated into, or influenced by, the developments at Angkor. Connections would have been required at the regional and supra-regional scales. Although there is unequivocal evidence that some of the temples were directly associated with the road systems, the majority of temples are not situated adjacent to known transportation routes.

In order to ‘connect’ the temples and reflexively conceptualise the development of the transportation system, we need to determine which temples were built, renovated and used during the reign of individual kings. In this approach, contemporaneously occupied temples can be integrated into a transport network that reflects the influence of kingly empires. The relationship between transportation and temples will be examined in three ways:

  1. by using the location of Khmer temples as a means of identifying potential routes of communication between temples and settlement areas (land- or river-based);
  2. by separating temporal sequences of temple construction on the landscape in order to assess when potential routes may have been established or most frequently used; and then
  3. by using the known mapped roads and transportation infrastructure as a means of explaining temple placement.

Temple selection

Of the 2,450 pre-Angkorian and Angkorian period sites recorded in Cambodia (Bruno Bruguier, personal communication, August 2005), 22 temples outside of Angkor were selected for this study (see Fig. 6.1). The initial criterion for limiting selection of sites was based on BP Groslier’s description of the Khmer heartland, which is associated with the zone of central cultural and political control:

Figure 6.1 Location of Khmer temples discussed in this study

Although the Angkor plain prevailed, everything points to the fact that the Khmer really felt at home between the edges of the Lakes and the Se Mun valley, and more generally between the triangle of the Lakes and Tonle Sap to the southwest and middle Mekong to the southeast (Groslier 1986[1973]:39).

From a transport perspective, it is most likely that sites within this heartland had some form of intra-regional communication and a route to the capital. Four sites established during the pre-Angkorian period (Vat Phu, Sambor Prei Kuk, Banteay Prei Nokor, Prasat Andet) were included, as many of the Angkorian kings rebuilt or erected inscriptions at these older temples. Communication routes to these areas may therefore pre-date the Angkor period. The criteria used to further narrow the number of temples considered in this study were based on a combination of the following:

  • Proximity to the known Angkorian road system. Temples located adjacent to known Angkorian roadways are most obviously included in this study. The majority of temples, however, are not connected to the mapped network.
  • Temple prominence, specifically, the size and location of the temple or temple complex. Only large ‘state’ temples or temple complexes were selected following the idea that larger centres have more politico-religious interactions with the capital. Location refers to where a temple is situated; in particular, temples built upon hills which would have provided focal points of religious worship for the people and kings of Angkor.
  • Evidence of royal connection between temples and kings. Presence of information about a temple’s involvement in the greater political workings of the empire. This includes epigraphic and architectural data about the king who founded the temple, who and when modified the built structure, and any inscriptions written by or directly datable to specific kings detailing their involvement with the site.

Inscriptions are perhaps the most critical source of information for linking the provincial loci of the Khmer empire to the capital. The idea of using inscriptions to draw maps of the extent of the Khmer empire under each king was first adopted by Parmentier (1916). The temporal information used here to construct new maps is based on summaries of the original translations (Briggs 1951; Snellgrove 2004; Jacques and Lafond 2004). As with the selection of sites, the number of inscriptions used to construct the database are limited to those dedicated by, or associated with, a specific king. It is argued that this information indicates either direct influence by the capital or some form of communication between the sites. While an inscription might not have been directly written by the king to whom it refers, it is sufficient evidence to consider the temple linked to the greater communication network and the royal political milieu at that time.

Architectural style provides the other critical evidence of a king’s involvement at a site. Dates for the temples are derived from a combination of the original art historical chronologies (Stern 1927; Parmentier 1939; Coral-Rémusat 1940; Briggs 1951; Boisselier 1966) and more recent summaries of the temples of the Khmer empire by Snellgrove (2004) and Jacques (Jacques and Lafond 2004). Ironically, the actual dates of these buildings are reliant on the inscriptions carved into their walls or on stela found in their enclosures.

Generating temple connectivity with GIS

The process of modelling potential transportation connections is greatly facilitated by integrating the historical information for each temple into a GIS platform (ArcGIS 9 and MapInfo). The value of GIS in addressing specific research questions is the relative ease with which comparisons between the temple data and a wide range of digital spatial data sets (topography, area of cultivable land, population density, etc.) can be made. To examine the relationship between temples and transport routes, the three operations undertaken through GIS are to:

  1. connect each of the temples associated with a particular king;
  2. create a maximum extent, or ‘communication corridor’, for each king by connecting the most distant temples of his reign; and
  3. identify all areas of overlap between kings. For this last step the successive communication corridors of each king are layered and new zones extracted for every overlap that occurred (Fig. 6.2).

Figure 6.2 Example of the generation of shared communication corridors

These overlapping corridors are then organised according to the number of kings (one to 15) who shared influence over the area. Once the communication corridors are created, they are compared with digitised layers of the mapped Angkorian roads, rivers and watersheds, which consist of the tributaries and water catchment of a river. Topographic features (contours) are not included in this preliminary study.

Transportation analysis

The analysis of transportation routes involves three primary steps. The first is to establish the relationship of Khmer temples through time. The second step is to compare the GIS-derived information with the mapped roads to determine where other potential transport routes may have existed. Given the fact that transport is also an active agent in structuring settlement, the third step is devoted to examining the role of communication on the location of two Angkorian temples, Koh Ker and Banteay Chhmar.

Sites, Kings and Corridors

The results for the number of kings ‘present’ at a particular site are illustrated in Table 6.1. By making direct connections between the temples of each king in the GIS, a series of communication corridors is generated that relates to both the regional focus of the king and the potential transport routes between sites. Fig. 6.3 illustrates the change in the location of communication corridors for each king which influenced more than three temples between the ninth to 13th centuries. The first significant feature is the general bias to the northeast of Angkor. The importance of this region has long been recognised from the histories described in the inscriptions: ‘Champassak (southern Laos), was to remain the mystical source of Angkorian power, the place erected by the Angkor kings for their pilgrimage, more important in a sense than the former patronage of the Fou-nan’ (Groslier 1986[1973]:39).

The second significant feature is the restricted size of royal territories until the reign of Suryavarman I in the early 11th century. At this time the focus expands significantly to the northwest and south of the Tonle Sap Lake. Interestingly, the areas influenced by subsequent kings largely mirror the 11th century occupation, which suggests that regional communication was formally established at this time.

Figure 6.3 Communication corridors controlled by Angkorian kings between the 9th and 13th centuries CE


King Reign Temples
Jayavarman II 770–835 4
Yasovarman I 889–900 5
Isanavarman II 923–928 3
Jayavarman IV 928–942 3*
Rajendravarman 944–968 6
Jayavarman V 968–1001 6
Udayadityavarman I 1001–1002 2
Suryavarman I 1010–1050 13
Udayadityavarman II 1050–1066 10
Harshavarman III 1066–1080 3
Jayavarman VI 1080–1107 4
Dharanindravarman 1107–1113 3
Suryavarman II 1113–1150 10
Yasovarman II 1150–1165 2
Jayavarman VII 1181–1219 12
Indravarman II 1219–1243 3
Jayavarman VIII 1243–1295 2

Table 6.1 Number of Khmer temples ‘influenced’ by each Angkorian king *Does not include Angkor

Shared corridors and mapped roads

The next stage in the analysis is to look at where and when corridors were commonly held by the Angkorian kings and how this information relates to mapped roads. Part of the GIS application, as discussed above, involved the identification of areas of overlap or commonly shared communication corridors. These communication corridors represent the shortest routes between two sites, which in effect mimic the role of roads to directly link two centres. These results are grouped into clusters of zones with one to five, six to ten, and 11 to 15 kings respectively (Fig. 6.4). Viewed temporally, these basic divisions provide some interesting clues about the expansion of the empire and where communication developed. Not surprisingly, the furthest areas from Angkor were controlled by the least number of kings, and almost exclusively by the pre-eminent rulers of the 11th to 13th centuries, Suryavarman I, Suryavarman II and Jayavarman VII.

By contrast, the most commonly shared region, located between Angkor, Preah Vihear and Neak Buos, was held continuously from Jayavarman II (ninth century CE) through to Jayavarman VII (13th century CE). The potential longevity of interaction in this corridor strongly supports the need for a communication route within the area, and there is indeed one recognised road located within this zone. This confirms that the most commonly occupied Angkorian territory contained a formalised communication route. The implication for the age of the roads, in particular the route connecting Angkor to Vat Phu, is that they predate the Preah Khan inscription of Jayavarman VII (1181 CE).

Figure 6.4 Location of commonly shared communication corridors and mapped Angkorian roads

Mapped roads, rivers and watersheds

Control of water has been a recurring feature in Southeast Asian culture since at least the early metal ages (Malleret 1959b; Moore 1989). Water management reaches its technical apogee during the Angkorian period with the construction of the great baray, embankments, canals, moats and bridges throughout the Khmer empire (Groslier 1979; Fletcher et al 2003). The concept that the Khmer actually controlled, or at least used, the rivers to facilitate movement has rarely been discussed in relation to transportation (Im 1998). Fig. 6.5 shows that the majority of Khmer temples are situated away from the main rivers (6.5a) or are located on the watershed (6.5b) between two catchments (Banteay Chhmar, Beng Mealea, Prasat Andet). Only Preah Khan of Kompong Svay and Banon, which is a hill temple, are situated on the main river. The reason for choosing to locate a site away from the main river may relate to avoidance of annual monsoon flooding, the tendency of large Cambodian rivers to meander, or the potential use of main rivers as a means of attacking settlements. The general settlement pattern, however, indicates that access to a natural waterway was a predominant factor in site placement.

Figure 6.5 Location of Khmer temples relative to rivers (a) and river catchments (b)


Figure 6.6 Comparison of the foundation date of Khmer temples and river catchments

The incorporation of river and watershed data is useful for clarifying the patterns of regional temple development discussed in relation to communication corridors. Comparing the position of temples and watersheds through time (Fig. 6.6) we see that the focus of Angkorian settlement for the first two centuries is the catchment of the Sen River, a fact previously noted by Hall (1985:173). It is not until the reign of Suryavarman I that other catchments, particularly the Nam Mun in northeast Thailand and Sangker around Battambang, become regularly incorporated into the empire.

An integrated transport network

From this discussion, both road and river networks appear to have facilitated communication between the Angkorian temples and settlements. The relationship between the roads and rivers, however, is only truly manifest when examined as an integrated network (Fig. 6.7). Without exception, roads bisect the space between the major rivers, running perpendicular to the flow of water in the watershed. While topography is not considered in this study, the shallow gradient across most of central Cambodia facilitates the development of these cross-cutting routes. These routes greatly expand the range of accessibility, both spatially and seasonally, in this region.

Figure 6.7 The terretrial and riverine communication networks during the Angkor period

Towards a local historical geography of transportation

The dynamic relationship between settlement and transportation can also be applied to the debate surrounding site selection. Two temples whose positions on the landscape are enigmatic in Khmer history are Koh Ker, the temporary capital associated with Jayavarman IV (928–942 CE), and Banteay Chhmar, one of the temple fortresses of Jayavarman VII (1181–1219). The elaborate structures built at these sites suggest they were of great importance, but the environment in which they are situated has challenged rational academic explanation. A closer inspection of these two case studies will reveal that the selection of the sites is more purposeful than previously thought.

Koh Ker

In 928 CE, Jayavarman IV seized control of the Khmer empire and established his capital away from Angkor at what is now the location of Koh Ker (Càdès 1968:114–115). Following Jayavarman IV’s death in 944 CE, the site was ‘abandoned’ and the capital re-established at Angkor. Visitors and scholars to the area have long remarked on the apparent lack of geographic or economic benefit for selecting Koh Ker over the rice basket and fish bowl of Angkor and the Tonle Sap Lake.

From the analysis of geography and transportation, four relevant points emerge that can shed light on this curious location (Fig. 6.8). First, Koh Ker is situated in the Sen River catchment and on a tributary that connects to a significant number of the larger Angkorian temples. Second, it lies on the edge of the most commonly shared communication corridor (11 to 15 kings) and specifically on the edge of the corridor held by 11 kings, dating from Bhavavarman II (late seventh century CE) to Jayavarman VII (mid-13th century CE). Third, part of the road connecting the Angkor area to Vat Phu is located in this corridor, just to the north of Koh Ker. And lastly, besides the Jayavarman VII hospital built at Prasat Andon Kuk, no other evidence of architectural construction (other than that of the inscriptions) has been found at Koh Ker which can be attributed to Angkorian kings after the death of Jayavarman IV. While there has been limited fieldwork at Koh Ker, it is possible that this ‘abandonment’ is related to local environmental limitations for continuous, intensive agriculture. The landscape around Angkor, specifically its proximity to the Tonle Sap Lake, provides a much more tenable basis for supporting large populations.

Figure 6.8 A transport geographic interpretation of Koh Ker

Using transportation as a guide, we can posit new interpretations for the selection of this site. Koh Ker is situated in the ‘mini-homeland’ of the Sen River valley which connects it via waterways to the major Angkorian sites in the region. More importantly, following Groslier’s comment about the significance of the Champassak (quoted above), Koh Ker is located next to the road that connects Angkor to the pilgrimage sites of Vat Phu, Preah Vihear and Neak Buos. In fact, Koh Ker’s close proximity to these religious sites would effectively cut off Angkor from the important religious centres, emphasising the significance of the new capital and isolating, and therefore diminishing, the importance of the old seat of power. The politico-religious implications of transportation may have overwhelmed the practical issue of sustainability in the area.

Banteay Chhmar

Like Koh Ker, scholars have long pondered the apparent isolation of the Banteay Chhmar temple complex. Even today travel to the site remains difficult and the surrounding landscape offers no obvious advantage to positioning a settlement in this region. Built by Jayavarman VII in the late 12th century as a tribute to his son, crown prince Indrakumara, who died during battles with the Cham (Càdès 1968:176), Banteay Chhmar is one of the largest and most impressive temples in the Angkorian empire. From a transportation perspective the site appears to be dissociated from both the Angkorian roads and the major river routes (Fig. 6.7). The former is particularly interesting given that, as Jayavarman VII was a major proponent of road infrastructure and the indisputable importance of the route between Angkor and Phimai, the only immediate evidence that Banteay Chhmar is part of a greater transport network is the appearance of a rest-house in the main enclosure. The communication corridors associated with the site are only shared by three kings, Suryavarman I, Udayadityavarman II and Jayavarman VII, showing that influence in the region was late.

Figure 6.9 A transport geographic interpretation of Banteay Chhmar

The use of riverine transport is also questionable, as the site is not situated directly on the nearest tributary. Recent research by Pottier (2004), however, has identified the remains of an extensive canal network radiating out from Banteay Chhmar. When the location of these canals is examined in relation to watersheds, it becomes apparent that these man-made features could have connected the site to both the Mongkol Borey and Stung Sreng drainage basins and integrated Banteay Chhmar into a greater Angkorian transport network (Fig. 6.9). This would have significant military and economic advantages. Positioning Banteay Chhmar between these two catchments served to facilitate access to both the Chao Phraya and Mun River valleys in central and northeast Thailand. Travel to Angkor could therefore be undertaken to the south or east by river to the site of Phnom Srok and the beginning of the Angkorian road that connects to the West Baray. The construction of these waterways to increase local and regional mobility is yet another example of Khmer water-control practices.

A temple landscape connected

The focus of this chapter has been to introduce the concept that the ‘isolated’ Khmer temples were part of a dynamic communication network. The mapping of potential routes provides new insights into the relationship of temples and their geographic setting and, perhaps more significantly, identifies the significant role of transportation in the history of their development. By comparison with the preconceived notions of Khmer transport history, the GIS model offered here suggests that the ‘classic’ distribution of sites and terrestrial transport infrastructure dates back to the mid 11th century during the reign of Suryavarman I. These routes, however, were likely being exploited before this date, a fact attested to by the concentration of sites located along water routes, such as the Sen, from early in the pre-Angkorian and Angkorian periods. A final facet of this transport approach is that the need for mobility and communication can provide significant insight into understanding why sites are located where they are.

This study presents preliminary ideas on the role of transportation; it was not possible, nor was it desirable from a purely logistical perspective, to consider all of the available information, as we are interested in the physical limitations to transport through time. At the very least, this chapter offers a new path for investigating interactions in the Southeast Asian past. The application of transport theory must continue to utilise remote sensing, GIS and ground-testing data as a means of developing and verifying models. These basic patterns of communication must be further examined in relation to topography, a more detailed analysis of the inscriptions, the location of road infrastructures (bridges, tanks, rest-houses), and the practicalities of river navigation.


Numerous people were involved in the process of translating this chapter from thoughts to reality. Eileen Lustig, Ngaire Richards and Damian Evans provided critical information, database set-up and GIS expertise. Critical editing and discussions were shared with Liz Holt, Dan Penny, Associate Professor Roland Fletcher, Dr Kyle Latinis, Martin Polkinghorne, Dr Christophe Pottier, and Andrew Wilson. I want to thank the Archaeological Computing Lab at the University of Sydney for providing access to the computer programs needed to undertake the GIS portion of this chapter. I am indebted to Larry Crissman at Griffith University who provided access to digital geographic resources. Lastly, I want to thank Dr Alexandra Haendel for organising the ‘Sacred sites’ conference in 2005 and inviting me to participate, and the Monash Asia Institute and Centre of Southeast Asian Studies for financial support for my attendance. This research was funded in part by the Social Sciences Humanities Research Council of Canada doctoral fellowship, the University of Sydney and the Australian Research Council. Any errors or omissions are mine alone.

Old Myths and New Approaches: Interpreting Ancient Religious Sites in Southeast Asia

   by Alexandra Haendel (editor)