Large Terrace Structure Unearthed in the Heart of the City Zone of Īśānapura: Could It Be the ‘Great Hall’ Described in the Book of Sui?

Abstract

The archaeological complex of Sambor Prei Kuk is identified with Īśānapura, the capital of the Chenla Kingdom in the early 7th century, a political predecessor to the Angkor Empire. Previous studies have largely focused on the eastern temple zone, where numerous Hindu religious structures are concentrated, while the western moated city area—presumably the core of urban and political activity—has remained underexplored. This paper presents the results of recent archaeological excavations at a large central mound within the city zone, which is hypothesized to have functioned as a key administrative facility. The excavation revealed a large-scale and uniquely configured terraced structure, unprecedented within the site, and radiocarbon dating suggests its construction dates from the mid 6th to mid 7th century. Notably, a Chinese historical source from the same period, the Book of Sui, describes a “Great Hall” in Īśānapura where the king administered state affairs. The characteristics and spatial context of the excavated structure suggest a possible correspondence with this account. In addition, LiDAR-derived topographic data identified multiple linear and structural features extending eastward from the mound, indicating that it may have stood at the rear of a more extensive administrative compound. These findings offer new insights into the urban structure of Īśānapura and the political architecture of early Khmer civilization.

1. Introduction

The political role played by the rulers at the apex of centralized power in the Angkor Empire was immeasurable. The king acted as the protector of the people, conferring titles, offices, and ranks upon his subjects, erecting temples to demonstrate the legitimacy of the throne, and constructing and maintaining water management facilities, as well as updating transportation infrastructure, establishing monasteries for education, and operating hospitals for healthcare. Furthermore, the king served as the champion of law (Dharma), the highest court judge in the kingdom, and the supreme commander of the military [1]. However, information regarding the aspects of rulers’ governmental activities, the facilities where these duties were carried out, and the locations within urban centers of these administrative buildings is limited.

Clearly identified structures believed to be royal palaces are limited to Banteay at Mahendraparvata, which is considered to be the capital from the late 8th to early 9th centuries, Prei Monti built in the Hariharālaya, believed to be the capital of the 9th century, and a section in the central northwest quadrant of Angkor Thom. Attempts to identify the locations of royal palaces or “throne halls” of successive kings in Yashodharapura, where the Angkor Empire was utilized as the capital over the long term, are not new [2]. Given that most palace facilities were likely wooden structures, their identification amidst the overlapping historical traces within the Angkor archaeological complex is challenging. However, identifying the state temples and royal palaces at the heart of cities provides significant clues to understanding the centers, scale, and composition of successive urban settlements.

Various approaches have been attempted to reconstruct the layout of royal palaces and associated facilities, including archaeological excavations on-site [3,4], interpretation of Chinese historical texts such as the Customs in Cambodia [5], analysis of wooden structures depicted in temple reliefs at sites like Bayon and Banteay Chhmar [6,7], and comparative analyses of palaces across Southeast Asia, including different periods [8].

In this context of palace studies, the 7th-century Chinese historical document “Book of Sui” has received little attention, despite containing descriptions of the activities of Chenla kings and the building known as the “Great Hall.” While it is already an established theory that the current Sambor Prei Kuk archaeological site corresponds to the Chenla capital of Īśānapura, there has been no archaeological debate on the existence or location of this structure within the site.

Recently, excavation surveys conducted at the large mound (Site M.90) located in the central area of the city zone revealed the discovery of a unique and large-scale structure unparalleled elsewhere in the site. Based on its location, size, and structure, the significance and uniqueness of this structure are evident, distinguishing it from the numerous temples found within the site.

This paper is structured as follows: Section 2 reviews previous research on the moated city zone of the Sambor Prei Kuk archaeological complex and highlights the presence of a group of structures concentrated from the city center to the eastern area. Section 3 presents descriptions from the Book of Sui regarding the city of Īśānapura and its administrative building, the so-called “Great Hall,” along with the methods of archaeological excavation. Section 4 outlines the excavated structures and the evidence used to estimate their dates of construction. Based on these findings, Section 5 offers a discussion on the original construction and later modifications of the structures, as well as the rulers who may have been involved. It also explores the layout of the presumed surrounding complex and the nature of the political and ritual governance system, drawing comparisons with later Angkorian urban and palace configurations.

2. Distribution of Archaeological Features in the Moated City Zone

The archaeological site of Sambor Prei Kuk, located in Kampong Thom Province, Cambodia (Figure 1), used to be known as Īśānapura, the capital of the Chenla Kingdom during the first half of the 7th century. The archaeological complex is comprised of the eastern temple zone where numerous temples are densely clustered, and a western city zone with an almost square layout spanning approximately 2 km on each side.

In the eastern temple zone, there are large temples including Prasat Sambor, Prasat Yeai Poan, and Prasat Tao, and statues of Hindu deities are believed to have been enshrined there. Multiple Khmer inscriptions have been found in these temples, and based on their content, Prasat Sambor and Prasat Yeai Poan are attributed to the reign of Īśānavarman I [9]. Meanwhile, the moated city zone is a 4 km² area surrounded by a linear moat on three sides excluding the east side, which is bordered by a seasonal river called O Krou Ke (Figure 2). The land inside the moated city is level, spreading at an altitude of 20 to 24 m and gradually dropping to an altitude of 14 m toward the river in the east.

Regarding the western city zone, structural features suggesting the existence of a city wall were first identified by Parmantier in the 1920s [10], and in 1937, an aerial survey conducted by Goloubew confirmed the presence of a moat [11,12]. However, comprehensive investigations were not carried out until the late 20th century. It was not until 1998 that extensive field surveys and interviews with local residents were initiated, leading to the confirmation of the presence of numerous structures.

Within the entire archaeological complex, there have been records of 173 sites and just under 300 brick structures to date [10,13,14,15,16,17,18,19]. Based on the investigation conducted by the authors until 2023, a total of 86 sites indicating the presence of brick and laterite structures, or terrain with a high likelihood of containing such structures underground, have been identified within the city zone [20]. Among these sites, 14 are believed to include multiple structures in close proximity, forming a single complex. When counting the structures individually, the total number amounts to 103 structures. Some structures are relatively well preserved, such as brick shrines M.61 (Pr. Tamon) and M.75 (Pr. Donmong), but many are in an advanced state of collapse, and most have been damaged by looting at the center of the structure. The remains within the city zone, while smaller in scale compared to the complex structures in the temple zone, include some complex facilities composed of multiple shrines, as well as some with surrounding walls and moats. Additionally, within the city zone, a total of over 150 reservoirs of various sizes have been identified.

In addition to the brick and laterite structures identified through previous surveys, several linear features were observed within the moated city zone [21]. However, more precise and reliable data have recently been obtained through airborne LiDAR surveys conducted by the Cambodian Archaeological LiDAR Initiative. In particular, the central area of the site is densely covered with forest, which limited the effectiveness of traditional aerial photography and ground surveys in detecting subsurface features. The application of LiDAR technology enabled the identification of several previously undetected earthworks in this heavily vegetated zone.

The use of LiDAR has proven highly effective in archaeological investigations in tropical rainforest environments, where dense vegetation often obscures surface features. Numerous studies have demonstrated its utility, with notable examples including research on Maya sites in Central America [22], as well as more recent advances in Cambodia.

The LiDAR data acquisition for this site is detailed in Evans, but an overview is provided here [23]. Airborne LiDAR data were collected using a Leica ALS70 HP sensor mounted on a helicopter flying at an altitude of 800–1000 m and a speed of approximately 80 knots, achieving a point density of over 16 points per square meter and a vertical root mean square error (RMSE) of approximately 15 cm. The sensor operated in MPiA mode with a 500 kHz pulse rate, a 45° scan angle, and 50% swath overlap. Positioning was supported by dual-frequency GNSS (L1/L2) and a Honeywell IMU, with differential corrections applied from a benchmark network (±20 mm horizontal, ±50 mm vertical accuracy). More than 100 RTK-surveyed ground control points were established per survey block. Data processing was carried out using Terrascan and ArcGIS. Vegetation filtering and ground classification followed established protocols, and a Digital Terrain Model (DTM) was produced for subsequent spatial analysis.

The LiDAR-derived topographic data revealed a series of linear features that divide the entire city into a 4 × 4 grid, suggesting the existence of a planned urban layout based on a grid system. At the intersection of the east–west and north–south central axes of this grid pattern lies the city’s midpoint, marked by M.173, a mound measuring 2 m in height. Presently, no bricks or laterite blocks are found on this mound or in its vicinity, and the existence of subterranean structures remains uncertain. Nevertheless, the presence of some structure or feature signifying the geographical center of the city is inferred.

From this mound, there is a relatively distinct linear feature extending eastward, and based on the results of subsurface exploration, it is inferred to be a water channel-like structure approximately 20 m in width [21]. This linear feature parallels the northern boundary of the surrounding wall centered around M.138 and further passes through the central area of a complex of facilities centered on M.75 (Figure 3). Although no traces of this feature are observed on the east side of the Ou Krou Ke River, extending this straight line eastward would lead to the central shrine of Prasat Sambor, serving as the city’s primary axis that traverses both the temple and city zones.

M.90 is located approximately 70 m south of this central axis and is the closest structure to M.173, which is situated at the center of the city. The area surrounding M.138, located to the east of M.90, is enclosed by a surrounding wall and it includes four brick-built shrines numbered as M.78 and M.79. However, M.138, situated in the center of this precinct, is a low terrace-like structure with an 8 m long square plan with projection to four sides and a height of about 1 m. Several pillar holes have been confirmed on its upper surface, indicating a different type of structure from the typical shrine, as revealed by an excavation survey [21]. Additionally, excavations at the central portion of the eastern perimeter wall did not reveal any structures resembling gates, suggesting a functional difference from conventional temple complexes.

From the eastern district, complex structures centered around Site M.75 were detected through LiDAR topographic data. (Figure 3). M.75 is a temple with three shrines located within the enclosure wall with the east gate, and the central shrine exhibits well-preserved elements of the Pre-Angkorian style. Surrounding this temple, another outer enclosing structure of approximately 235 m is evident. Positioned on the central east-west axis of the outer wall is a brick structure, identified as M.172, believed to be the eastern gate of this outer wall. In the north area within this outer wall, there are two brick structures (M.72) and one other brick structure (M.73), and outside the outer wall, in close proximity, several brick or laterite structures (M.70, M.71, M.74, M.78, M.77, and M.80). These structures are considered to have formed a complex of facilities in the front area of the urban center.

These complex facilities located to the east of M.90 contain temple-like structures within the premises; however, their layout differs from the temple compounds in the temple zone, suggesting that they may have served purposes other than religious activities.

3. Textual Materials and Methods

3.1. Īśānapura as Described in the Book of Sui

Regarding Īśānapura, some information can be gleaned from Khmer inscriptions and Chinese historical records. To date, a total of 25 Khmer inscriptions have been discovered in the Sambor Prei Kuk archaeological complex, when considering the Robang Romeas temple to the north to be part of the complex. Among these inscriptions, five are from the 10th to 11th centuries, while the rest are considered to date back to the 7th century, with at least seven being associated with the reign of Īśānavarman I [9,24]. The information obtained from these inscriptions primarily focuses on religious matters, such as the installation of deity images in shrines and donations to temples, as well as descriptions of the king’s achievements and praises. Information about general societal conditions, common beliefs, and urban structures is limited.

Chinese historical records offer valuable insights into foreign relations, urban planning, political governance, and cultural practices, presenting a perspective distinct from Khmer inscriptions. The primary source documenting Īśānapura is found in the Book of Sui, one of the Twenty-Four Histories, authoritative Chinese texts chronicling the histories of successive Chinese dynasties. Published in 656 AD, the account of Īśānapura in the Book of Sui (Biography of Southern Barbarian, Volume 82) stands out as one of the most comprehensive among Chinese dynastic historical texts. Subsequent texts, such as the “Bei Shi” (Book of the Northern Dynasties, Volume 95), “Jiu Tang Shu” (Old Book of Tang, Volume 197), “Taiping Yulan” (Volume 786), “Tongdian” (Volume 188), “Xin Tang Shu” (New Book of Tang, Volume 222 Part 2), “Tang Huiyao” (Volume 98), and “Taiping Huanyu Ji” (Volume 177), among others, also contain similar accounts of Īśānapura [24].

The richness of historical accounts during this period likely stemmed from the concurrent rise of the emergent Chenla Kingdom, marking a shift in power from Funan, which garnered considerable attention in China as a significant upheaval in the Southeast Asian region. Moreover, during the early 7th century of the Sui dynasty, there was a growing interest in the South China Sea region, prompting Emperor Yang of Sui to dispatch envoys to various Southeast Asian territories in the third year of the Daye era (607) [25].

The kingdoms mentioned in the Book of Sui regarding the South Seas include Chenla along with Linyi (central Vietnam), Chitu (southern Malay Peninsula to western Java), and Bali (eastern Java). This limited mention suggests the prominence of Chenla as a powerful entity during that time. Additionally, tributes from Chenla to China were frequent, and Chinese historical records confirm nine instances of tribute during the 7th century in the years 616, 623, 625, 627–649, 628, 635, 651, 682, and 698.

The description of Īśānapura in the Book of Sui covers a wide range of topics, including its location, the city, the king’s governance, government buildings, society, customs, geography, climate, and religion. For the purpose of this paper, the English translation of relevant extracts from the Book of Sui, related to the city’s structure and government buildings, are provided below.

[omitted text before]

In the royal city of Īśānapura, there are over 20,000 households. In the center, there is a great hall from where the king governs the realm.

[text omitted]

The king presides over the affairs of state every three days, seated on a pedestal made of fragrant wood and adorned with the Seven Treasures. Above his pedestal hangs a curtain of fine cloth, supported by decorated wooden poles, enclosed by walls adorned with ivory and precious metals, resembling a pavilion that gleams with gold and light.

[text omitted]

Those who approach the king must first touch their foreheads to the ground at the lower steps. If the king summons them to the upper level, they kneel with arms crossed. They sit in a circle around the king to discuss matters of state, and afterward, they kneel and prostrate themselves before departing. In the lower courtyard and at the gates, there are over a thousand guards wearing armor and wielding spears.

[text omitted]

Every year, during the fifth or sixth month when illnesses become prevalent, white pigs, white cattle, and white sheep are [sacrificially] offered outside the western gate of the city.

[text omitted]

To the east of the city, there is a deity named bha-ta-ri [Badreśvara]. In the rituals, they use human flesh. The king sacrifices a person every year and conducts prayers at night. There are a thousand guards present. This deity [Badreśvara] is honored in this manner.

[text omitted]

Many follow the teachings of Buddhism, while more believe in Dōshi [Hinduism], and temples housing idols from both Buddhism and Hinduism stand side by side.

[omitted text after]

The Book of Sui provides a clear account of the spatial arrangement in Īśānapura, with the western urban area from where the king governed and the eastern temple district. This description aligns with the layout observed during on-site investigations of the Sambor Prei Kuk archaeological site. Furthermore, the existence of the western gate of the city is mentioned, and on-site evidence, such as a substantial mound structure (M.107) with abundant fallen brick material at the central western edge of the city, corresponds to this description in the historical source. This reinforces the credibility of the account in the historical source concerning the spatial layout. Of particular interest is the “Great Hall” where the king conducted his political affairs. The king was situated on a raised platform higher than his officials, and the officials ascended and descended using a staircase. Furthermore, there was a spacious area around the king where multiple officials could be seated.

3.2. Methods and Scope of Archaeological Excavation Survey

Excavation surveys were conducted at the M.90 site in 2015 and 2022. Following established conventions from previous investigations at this archaeological complex, a 1-m grid system was laid out over the entire mound using a total station, and excavation areas were expanded systematically based on this grid. The overall orientation of the square-shaped city is slightly rotated counterclockwise, a pattern that is also observed in the city’s primary axis. Major temples at the site, believed to have been constructed in the 7th century, similarly follow a counterclockwise orientation. This characteristic persisted into the Angkor period, with the majority of temples, with only a few exceptions, exhibiting the same alignment. The north–south axis of the mound under investigation is also rotated approximately 11 degrees counterclockwise in alignment with this broader pattern. Based on the assumption that subsurface structures followed this orientation, the excavation grid was set to match the same angle.

The mound, with an overall area of about 2000 square meters, has been excavated within an area of 353 square meters thus far (Figure 4). The excavation primarily focused on the removal of sediment and collapsed bricks to confirm the presence of structural remains. In two sections (55R-S and 63U-V), the excavation went deeper beneath the upper floor of the terrace to examine the internal structure of its platform. It should be noted that while excavation in plot 55R-S reached the ground level surrounding the terrace, it did not reach the natural soil layer. Architectural features and soil strata exposed during the excavation were recorded through photogrammetry, using both drone and handheld camera imagery. These photogrammetric records were subsequently used as base maps for detailed field drawings created through direct measurement on site.

4. Results

4.1. Structures Confirmed in the Excavation Surveys

As a result of the excavation survey, remains comprising laterite and brick, with a plan dimension of approximately 70 m north to south and 15 m east to west, were detected. These remains consist of a long terrace-like structure extending north to south, a laterite-built elongated chamber attached to its southern side, and multiple structures made of brick and laterite masonry on the surface of the terrace-like structure The structures on the terrace appear to be additions made in later years. While many of the brick-built structures identified within the archaeological site have typically been associated with temple components such as shrines and gates, this structure stands out as unique in its design and scale.

4.1.1. Central Terrace

A terrace-shaped structure 47 m from north to south and 15 m from east to west was confirmed surrounded by a laterite masonry wall 2.1 m high (Figure 4). It has a planar shape that protrudes outward by two tiers on the north side and one tier on the south side. The laterite blocks have various sizes, but they are carefully stacked and flattened on their outside to a wall thickness of 0.7 to 0.8 m (Figure 5A). Based on the excavations conducted in limited areas on the northern and northwestern sides, the stair structure of this terrace has not been confirmed.

When a partial excavation was conducted within the terrace, it was confirmed that the interior of the terrace was filled with sandy soil containing a large number of earthenware fragments. No evidence of a laterite or brick-paved floor surface was found on the terrace’s upper level. However, based on the stratigraphic layer, it was inferred that a floor surface existed at a height of approximately 1.5 m above the surrounding ground level. Furthermore, several additional structures made of brick masonry were confirmed on the terrace structure, as will be described in detail later.

Around the terrace structure, layers of fallen bricks were excavated. In the excavations near the central part of the eastern face, upon removing this thick layer of partially collapsed masonry materials, it was observed that the upper layer of the laterite wall had collapsed outward (Figure 5B). Due to the substantial amount of fallen bricks in the vicinity, it is also estimated that there might have been brick walls in the upper layer of the laterite wall.

Figure 5. Excavated structures in each area. (A) Excavated laterite retaining wall of the Central Terrace, view from west (grid: L, 68–69). (B) Collapsed laterite and brick materials east side of the laterite wall of Central Terrace, view from northeast (grid: Z-e, 62–63). (C) Connection between south wall of Central Terrace and Southern Attached Chamber, view from southeast. (D) Excavated structure of the Southern Attached Chamber, view from south. (E) Excavated additional brick-built features on the Central Terrace, view from south. (F) Excavated additional brick-built features on the Central Terrace, view from northwest. (G) Collapsed brick-built features on the Central Terrace, view from east (grid: O-S, 58–63). (H) North wall of the Central Terrace and brick-built features on the laterite retaining wall, view from north.

Figure 5. Excavated structures in each area. (A) Excavated laterite retaining wall of the Central Terrace, view from west (grid: L, 68–69). (B) Collapsed laterite and brick materials east side of the laterite wall of Central Terrace, view from northeast (grid: Z-e, 62–63). (C) Connection between south wall of Central Terrace and Southern Attached Chamber, view from southeast. (D) Excavated structure of the Southern Attached Chamber, view from south. (E) Excavated additional brick-built features on the Central Terrace, view from south. (F) Excavated additional brick-built features on the Central Terrace, view from northwest. (G) Collapsed brick-built features on the Central Terrace, view from east (grid: O-S, 58–63). (H) North wall of the Central Terrace and brick-built features on the laterite retaining wall, view from north.

4.1.2. Southern Attached Chamber

A long and narrow laterite chamber measuring 19 m in the north–south direction and 4.2 m in the east–west direction was confirmed attached to the south side of the central terrace (Figure 4 and Figure 5C,D). The inside space measures 18 m in the north–south direction and 2.4 m in the east–west direction. No door opening for entering and exiting the chamber has been confirmed in the surveys carried out to date. On the inner surface of the western wall, there are regularly spaced recesses vertically, with a width and depth of approximately 0.5 m, suggesting that wooden square posts were inserted to support a wooden roof. It should be noted that no tiles were excavated in this area.

The presence of vertical joints in the stone masonry between the south wall of the central terrace and the wall of the southern attached chamber suggests a difference in the timing of construction, hinting at the possibility that the central terrace was constructed first. However, since the size and masonry technique of the laterite used in the two structures are consistent, it would be reasonable to assume that the southern attached chamber was built shortly after the construction of the central terrace.

4.1.3. Structures on the Central Terrace

Multiple small-scale brick and laterite structures were excavated on the central terrace. These structures are roughly constructed using reemployment of materials, and they are situated on top of the thin layer of accumulated soil on the terrace surface. This suggests that they were added after a certain period following the initial construction of the structure.

Upper Low Terrace

On the north part of the central terrace, another elevated structure was confirmed, measuring approximately 4.3 m wide in the east–west direction, 14 m long in the north–south direction, and 0.3 m high (Figure 4 and Figure 5E,F).

Upper Brick Square Bases

Square bases made of brick, measuring 1.3 m on each side are arranged systematically at 1.5 m intervals from north to south on the east and west sides of the abovementioned upper additional terrace (Figure 4 and Figure 5E,F). The brickwork is rough, and the state of preservation is poor, with a maximum of six layers of bricks remaining. From the scope of the survey conducted to date, there appear to have been ten such bases from the north end of the terrace, but there is a possibility that they continue further southward to fill the entire space on the terrace. From their systematic arrangement, they may have been the bases of wooden posts, if so, it means that a large framework had existed with a 7 m-long beam spanning it in the east–west direction. Fragments of tiles have been unearthed from the terrace surface, but the count is only 10 pieces, and there is no definitive evidence suggesting that tiles covered this conjectured large roofing structure.

Collapsed Brick-Built Features

From the central excavation area of the central terrace (Grid 58–63, O-S), traces of a partially collapsed masonry structure were identified (Figure 5G). The remains had fallen from the northwest to the southeast, and it appears that at least 31 layers of brick were stacked, indicating that the original structure was over 2 m in height. However, the exact original form is uncertain. Based on the characteristics of the bricks, it is considered to date from the same period as the aforementioned lower terraces and square bases found on the terrace surface.

Upper Structures at the North End of the Central Terrace

Multiple low structures were confirmed at the north end of the central terrace. However, their original appearance cannot be estimated as the scope of the survey was limited, and because their remaining condition is poor and they were crudely made to begin with. On the protrusion on the north side, bricks are directly stacked on top of a laterite retaining wall, and a four-step stair-like structure was confirmed leading down to the north side (Figure 5H). This structure is rather well made compared to the brick masonry on other terraces, so it may have been built at the time of the initial construction.

4.2. Unearthed Artifacts That Contribute to the Dating of Structures

The excavation surveys unearthed more than 6700 potsherds. Most of these are earthenware fragments and more than 4000 were unearthed from the trench that was dug where the central terrace and southern attached chamber connect with each other. Many of these were found in the accumulated soil immediately outside the walls of the terrace. It is considered that the considerable number of artifacts in this soil originated from a period when the terrace was in use, although its maintenance was somewhat irregular. In grids 55R-S (down to a depth of 2.5 m from the estimated original top surface of the terrace) and 63U-V (down to a depth of 1.7 m from the estimated original top surface of the terrace) on the central terrace, the artificial fill of the terrace was surveyed by digging down into it, and as a result, numerous earthenware fragments were unearthed mixed in with the artificial fill.

The breakdown of the excavated artifacts includes 6660 pieces of earthenware or clay products, 10 pieces of Khmer or imported ceramics, 32 pieces of stone or stone products, 4 metal objects (of which 3 are iron), and 21 clay clumps. Among the earthenware fragments, the identified types are limited but include 101 jars, 44 large jars, 48 spouted vessels (kundis), 286 bowls, 7 tall jars, 1 lid, 1 small bowl with many holes, 8 stoves, 10 tiles, 3 small containers, and 3 disc-shaped clay items (Appendix A). It is worth noting that the tiles are all flat tiles, with two types identified: those with small holes and those with protrusions.

At the archaeological site of Sambor Prei Kuk, the types and shapes of earthenware have been classified in past archaeological surveys, but they have not been able to be chronologized, and their dates have not been determined [26,27,28,29]. Most of the unearthed artifacts were identified as fragments of jars and bowls, but they also include fragments of spouted ware, miniature earthenware, clay discs, stoves, and roof tiles. Some of the jar fragments are decorated with incisions and impressions on the neck and shoulder, and colored with red or white painting. All of the above are fragments of the common type and shape as those that have already been unearthed from this archaeological site to date and display no special types or features.

Below is a discussion of the decorative stone elements and Chinese and Vietnamese pottery that prove significant in chronologizing the construction and modification of the excavated structures.

4.2.1. Stone Items

Multiple sandstone fragments were unearthed, but only four display a specific shape or decoration. Three of them were unearthed from the accumulated deposit at the north end of the central terrace (grid X-Y, 77–81). These include two stone fragments (Figure 6A,B) with decorative carvings and a slightly larger, unidentified processed stone fragment (Figure 6C). The fourth fragment was unearthed from the accumulated deposit on the west side of the central terrace (grid L68). It is a round stone element thought to have been used as a stone axe or grinding stone (Figure 6D).

All of them were excavated above the floor level of the terrace and are items that fell to the soil not long after the terrace was abandoned. Regarding the two stone artifacts with decorative carvings, they are classified as of pre-Angkorian style, but their original form and purpose are uncertain.

4.2.2. Chinese and Vietnamese Potteries

Two sets of Chinese and Vietnamese pottery fragments were confirmed in the excavation survey and identified the date. One set of Chinese pottery fragments which were unearthed from the accumulated deposit in grid W69 on the central terrace would be parts of a single bowl. These fragments were excavated from the shallowest stratigraphic layer, at a depth of 5 to 20 cm, above the upper brick square base (Figure 6E). It is a Fujian/Guangdong-style blue-and-white bowl dated to the late 17th century.

The other was collected from the surface of the southern attached chamber (grid S25). It is part of a white porcelain bowl from northern Vietnam (Figure 6F). It is identified as from the 17th century. Similar artifacts have been found in Vientiane, Laos, and as far away as Kyushu, Japan [30,31].

4.3. Results of a Radiocarbon Dating of Sampled Charcoal

Charcoal samples were collected by digging down into the artificial fill inside the central terrace in grid 63U-V, and four of these samples were subjected to radiocarbon dating and tree species identification by accelerator mass spectrometry (AMS). The stratigraphic layers of the central terrace are divided into five major layers. The fifth layer was the internal fill during the construction of the central terrace, while the fourth layer is inferred to be the stratigraphic layer that was excavated from the original terrace to serve as the partial stabilizing foundation during the addition of the square brick base (Figure 7). Of the four charcoal samples that were dated, three were collected from layer 5 and one from layer 4.

The four charcoal samples were missing their final growth ring. This meant that there had been more growth rings on the outer side of the present state of the samples, so it needed to be assumed that the tree had died or had been cut down much after the date obtained from dating. With regard to tree species, one sample was found to be a carbide material from a broadleaf tree, but no further classification was able to be discerned. The other three were identified as three different tree genera; Lumitzera (Combretaceae), Dalbergia (Legunimosae), and Dipterocarpus. When the result of calendar-year calibration of each sample was observed within a two-sigma range (95.45% probability), the three charcoal samples collected from layer 5 indicated the period from the mid 6th century to the early 7th century, and the sample collected from layer 4 indicated the period from the latter half of the 6th century to the mid 7th century.

5. Discussion

5.1. Architectural Reconstruction and Dating of the Construction and Modifications of Large Terrace M.90

In the following discussion, the validity of identifying the excavated structure at M.90, which is situated near the center of the moated city, with the “Great Hall” described in the Book of Sui is examined. The analysis is based on the architectural features of the large terrace as reconstructed through excavation, its consistency with the historical description, and the estimated periods of construction and use.

5.1.1. Reconstruction of the Excavated Structure

A partial excavation survey conducted of the mound M.90 in the near-center of the moated city revealed a structure extending as much as 70 m from north to south as the largest structure within this city zone. It had been a unique structure, with a central terrace supported by a 2.1 m-high laterite wall and north–south long, narrow chambers attached to it on its south side (Figure 8). While excavation surveys conducted thus far have been partial, and stairs have not been confirmed on the eastern side, it is presumed that this terrace faces east, given that many temples within the archaeological site are oriented towards the east. Therefore, it is speculated that stairs may be located centrally on the eastern side. The reconstructed image shown in Figure 8 includes hypothetical elements such as the estimated height of the upper structure and the shape of the roof. These aspects are not directly evidenced by excavation and should be understood as interpretative hypotheses based on the arrangement of structural remains observed on the terrace.

At the northern end of the terrace, rectangular low terraces were arranged, and two rows of square bases were observed. All of these upper structures were made of reemployed brick and were constructed with rough brickwork. It is hypothesized that wooden columns stood on these square bases to support an upper structure. Within the excavation area, it has been confirmed that 10 columns extend in the north–south direction. If columns were evenly spaced along the entire length of the terrace, it would result in a colonnade structure with a total of 15 columns, suggesting a grand façade of this upper structure. As mentioned earlier, due to the very limited amount of excavated tile materials, it is unlikely that the roof of this building was originally tiled. However, tiles can be reused materials, and the possibility that they were transported in a later period should also be considered.

The chamber connected to the south appears to have been a long and narrow space with a width of 2.4 m and a length of approximately 19 m, enclosed by laterite walls up to the terrace’s floor level. Access from this chamber to the terrace was not possible, suggesting it may have been an auxiliary structure associated with the terrace. While no openings such as doors or windows for light and entrance have been confirmed so far, the possibility of such openings on the unexplored east wall is conjectured. There are recesses in the inner wall, which suggest that wooden columns were placed there, supporting a roof above.

The Book of Sui describes the throne within the Great Hall as being situated in a small pavilion atop a high platform, surrounded by several high-ranking officials, with subjects ascending and descending the platform via stairs. The excavated terrace structure fulfills all these descriptions, aligning closely with the depiction of the Great Hall in the Book of Sui. Additionally, both within and surrounding the Sambor Prei Kuk archaeological complex, a large mound structure designated as M.127, measuring approximately 65 m east to west, 30 m north to south, and 2.5 m in height, has been identified about 6 km north of the urban area. The mound exhibits scattered bricks and laterite on its surface, with some sections showing exposed laterite alignments, comparable in scale to M.90. However, due to its distance from the city zone and the absence of surrounding structures, it is implausible to consider this structure as the king’s administrative center. Thus, no other structure matching the description of the Great Hall in the Book of Sui has been identified as a viable alternative to M.90.

5.1.2. Chronological Consideration of the Construction, Modification, and Abandonment of the Excavated Structure

The charcoal samples collected from the fill soil inside M.90’s terrace (Layer 5), which would indicate the construction period of M.90, dated back to the mid 6th to the early 7th century. Furthermore, one sample collected from the soil layer related to the construction of the square base (Layer 4), which would indicate the period of addition when brick structures were added to the terrace, yielded results indicating a period from the late 6th to the mid 7th century. While these dates represent only the upper limits of the fill dates, the fill soil inside the platform contained numerous pottery sherds, suggesting that the soil used for filling may have been sourced from the surface soil of the surrounding residential areas. Therefore, it is unlikely that the dates indicated by these charcoal samples significantly predate the fill period.

The first construction period of the terrace slightly predates the reign of Īśānavarman I (616–635?), the most likely candidate responsible for the construction of Īśānapura. The results also include dates from the era of Bhavavarman I (580–600?), who initiated the expansion of Chenla’s power, as well as Mahendravarman (600–616?), and potentially even earlier periods.

In previous investigations in the city zone, results suggesting the possibility of the surrounding area being used before the 6th century have been confirmed. In excavation surveys conducted at 21 locations within the city zone in 2012, we identified artifact-bearing layers, mainly consisting of pottery fragments, in 17 out of the 21 locations, indicating that a wide area was utilized for some purpose [34]. Radiocarbon dating of charcoal samples extracted from the lower layers of artifact-bearing layer generally yielded dates ranging from the early 5th century to the late 6th century, while samples from upper layers indicated dates from the late 13th to the early 15th centuries, confirming long-term utilization of this city. Furthermore, radiocarbon dating of charcoal samples from beneath the structures and artifact-bearing layers at the M.49 and M.138 sites within the city zone predominantly yielded results dating back to before the 6th century. These findings indicate human activity in the region predating the recognition of this area as the capital of Chenla prior to the 7th century, supported by multiple lines of evidence. If the construction of the initial stage of this terrace structure dates back to the late 6th century, it suggests an era preceding the reign of Īśānavarman I, who is believed to have played a significant role in the construction of this city, implying the region’s political significance even before his reign.

On the other sides, the radiocarbon dates from charcoal samples in the fourth layer, which indicates the period when brick structures were added on the terrace, correspond to the reigns of Mahendravarman, Īśānavarman I, and Bhavavarman II, who are suggested to have established this site as a base. The temple constructions with surviving upper structures within the city zone such as M.61 (Pr. Tamon) and M.75 (Pr. Donmong), clearly exhibit pre-Angkorian styles contemporaneous with those found in the temple zone. Moreover, numerous examples of decorated lintels in the Sambor Prei Kuk style or Prei Khmeng style, as well as pedestals indicating pre-Angkorian forms, have been extensively documented from other collapsed structures [29]. This suggests that the construction of many temples within the city zone and the expansion onto terraces likely occurred simultaneously. In this way, the expansion period of the terrace structures in the early 7th century, as indicated by the dating results, coincides with a significant period of city construction, aligning with the dating provided in the Book of Sui regarding the “Great Hall.”

Even after the cessation of its use as the royal capital of Īśānapura, traces of continuous or intermittent utilization of this archaeological site have been confirmed through various remnants. Excavations in the temple district have revealed inscriptions and sculptures from the Angkorian period, along with traces of modifications in several structures, indicating significant alterations during the 10th to 11th centuries [27,28,29,35,36]. Additionally, discoveries of later sculptures, Khmer ceramics, and Chinese pottery further attest to its prolonged use well into the late Angkorian period and beyond. While previous excavations within the urban precincts have yielded artifacts from the Angkorian era, no Khmer ceramics or imported pottery from the Angkorian period have been unearthed from this terrace structure, suggesting the absence of evidence for its use during the Angkorian period at present. However, the presence of Chinese and Vietnamese ceramics from the 17th century, albeit in small quantities, suggests that the site was utilized during the post-Angkorian period. During this era, as the center of the Angkor Empire shifted southward, it is speculated that the urban area of Īśānapura regained some degree of significance.

5.2. Exploring Khmer Urban History Through the Great Hall of Īśānapura

Recent archaeological investigations at multiple urban sites of ancient and medieval Khmer civilization have significantly advanced discussions on the long-term historical development of Khmer urbanism [37]. While identifying royal palaces and administrative facilities—likely constructed of wood—in successive capital cities and regional centers remains a challenging task, new insights have begun to emerge from several key sites. In this section, the spatial planning of later Khmer cities is compared with that of Īśānapura to infer the characteristics of its urban layout and the underlying ideologies of governance and religious belief. Furthermore, by comparing the known spatial configurations of royal palaces in later periods with the archaeological traces within Īśānapura, this section explores the possible extent and composition of the royal palace complex that may have surrounded the “Great Hall.”

5.2.1. Characteristics of Īśānapura in Comparison with Urban Layouts of Later Khmer Cities

Khmer cities are broadly categorized into closed and open structures, with some surrounded by moats or walls while others lack such external fortifications. Closed cities may have either rectangular or irregular layouts. While there is debate regarding the definition and scope of what constitutes a city, we will focus on the major cities of the Khmer Empire for analysis.

Closed cities with rectangular or square outer walls are relatively rare, with examples including Īśānapura, Śresthapura, Banteay Prei Nokor, Phimai, Preah Khan in Kompong Svay, and Angkor Thom, among others. While exceptions exist, temples are often located at the center of square cities. In Angkor Thom, the Bayon Temple is located at the center, with the royal palace district situated to its northwest. While not based on a square layout, other enclosed cities include Angkor Borei and Longvek. At Angkor Borei, an elite walled district is assumed to have existed near the center of the outer enclosure [38]. Open cities, more common in Khmer capitals, often include areas or remnants believed to be royal palace districts or structures. At Mahendraparvata, Banteay is considered the site of the royal palace [39], while at Hariharālaya, Prei Monti is believed to be the royal palace site [40,41] (Figure 9). Additionally, at Chok Gargyar, although there is no conclusive material evidence, Andong Preng is regarded as the royal palace site [14,42,43]. Before the formation of Angkor Thom in the late 12th century, multiple phases of urban planning are believed to have taken place [2]. Additionally, excavations in the royal palace area of Angkor Thom have revealed multiple phases of construction, with recent research suggesting that its earliest phase dates back to the founding of Yaśodarapura [3,4].

It is not difficult to infer that the arrangement of the royal capital’s plan was centered around the placement of the state temple, known as the mountain temple. Alongside the temple mountain, the royal palace was presumed to be another significant element in urban planning. Jacques previously examined the urban layout of Yaśodarapura and proposed a city planning model where the royal palace was situated to the north of the temple mountain [2]. However, it has been noted that the application of Jacques’ model is somewhat limited, as the royal palaces in Hariharālaya and Chok Gargyar are positioned to the south of the temple mountains [44]. Nonetheless, the common feature of temple mountains and royal palaces being positioned north–south is evident across these cities.

In contrast, in Īśānapura, the temple district is situated to the east and the royal city district to the west, with the temple mountain Prasat Sambor and the central city extending along an east–west axis. This east–west orientation aligns with the description in the Book of Sui, which states “the god is to the east of the city.” Excavation surveys conducted in the vicinity south of Prasat Sambor, the temple mountain, have unearthed a significant amount of artifacts, including pottery and imported ceramics, suggesting it was an important area [29,45]. However, there is no clear evidence of structures resembling the Great Hall mentioned in the Book of Sui, and no definitive proof of a royal palace adjacent to the temple has been found. In other words, Īśānapura’s layout appears to differ significantly from the customary arrangement seen in later Khmer cities, where the temple mountain and royal palace are typically placed north to south.

What, then, do these spatial differences signify? While temple mountains served as centers of religious symbolism and ritual, royal palaces functioned as venues for political ceremony and administration. As such, they embodied two distinct symbolic cores within the city—divine and royal. The eastward orientation of Khmer capitals appears to have been consistent from the pre-Angkorian through the Angkorian periods. However, in Īśānapura, a pre-Angkorian capital, the temple district responsible for religious rituals was placed in the eastern front of the city, whereas the royal precinct handling governance and political affairs was located at the western rear.

In contrast, in early Angkorian capitals such as Mahendraparvata and Hariharālaya, the palace and the temple mountain were arranged in a generally north–south relationship. Yet closer examination reveals that the temple mountain was positioned slightly in front of the palace, closer to the city’s eastern axis. This spatial configuration evolved further in Yaśodharapura and later capitals, where the temple mountain and royal palace came to be aligned more strictly along a north–south axis. At major temple mountains such as Phnom Bakheng and Baphuon, the royal palace was placed directly on the same longitudinal axis as the temple (Figure 8).

Such a transformation in spatial planning may reflect a deeper ideological shift in the perceived relationship between kingship and divinity. In earlier periods, the king ruled as a representative or intermediary of the gods, borrowing divine authority to legitimize his rule. Over time, however, this relationship evolved into the Devarāja concept, in which the king came to be regarded as a divine being in his own right. This shift toward a more integrated sacral kingship may have been physically manifested in the spatial logic of urban planning, where the alignment of temple and palace increasingly symbolized the unity of religious and royal power.

5.2.2. The Spatial Position of the Great Hall Within the Royal Palace Complex

Based on the known layout of royal palaces in the Khmer context, let us consider the positioning of the structure presumed to be the Great Hall (M.90) within the palace complex of Īśānapura. It is generally assumed that royal palaces consisted of two main areas: the king’s private residential quarters and public administrative spaces [8], with the audience hall presumed to have been the central facility within the latter. Among these, the Great Hall served as the assembly hall where the king conducted his official duties and where administrative functions overseen by high officials took place. In the section “Entry and Exit of the Provincial Lords” in the Chinese text the Customs in Cambodia, it is noted that the king would sit in the government office twice daily to attend to state affairs, suggesting that the Great Hall could have served as a venue for such routine administrative duties. However, being a terraced structure, the Great Hall likely served as a facility intended for the display of royal authority. It is conceivable that it was used for various ceremonies and events related to state governance, including ceremonies attended by high officials and subjects, annual rituals, inauguration ceremonies, military audiences, as well as occasions for audiences with envoys, traders, Brahmins, and diplomatic relations.

In the central part of Īśānapura, traces of the enclosure wall of M.90 and adjacent structures are not currently visible on the ground, but the audience hall should be considered as one of the complex facilities. Known royal palaces of the Angkor period, namely Banteay, Prei Monti, Andong Preng, and those within Angkor Thom, all formed complex compounds. Banteay, for instance, is enclosed by an embankment measuring approximately 615 m east–west and 400 m north–south, with an eastern projection measuring 220 m east–west and 120 m north–south [39]. Prei Monti, situated within an 800 m east–west and 500 m north–south moat and enclosure wall, revealed multiple structures and residential traces during excavation [41]. Andong Preng, surrounded by dual laterite rows intermittently along its 240 m east–west and 188 m north–south dimensions, likely featured wooden galleries. Several laterite rows remain inside, indicating the presence of numerous wooden structures erected upon them [43].

The royal palace within Angkor Thom is enclosed by walls and moats, encompassing an area of approximately 585 m east–west and 245 m north–south. Internally, it would be divided into five distinct zones [46]. Numerous structures are visible above ground, and subterranean remains of precursor structures spanning several periods have also been identified [3,4,46]. Notably, in comparison to M.90, the Elephant Terrace located on the eastern side facing the Royal Plaza stands out. Stretching over 300 m north–south and with heights ranging from 2.9 m to 4.6 m, it boasts impressive dimensions, adorned with intricate carvings on its facade. This terrace once featured wooden pavilions where the king would observe events held in the Royal Plaza [47]. The existing Elephant Terrace dates back to the late 12th century, but prior to that, the presence of some structure in front of the eastern gate of the royal palace has been estimated [46,47]. The placement of a facility for the king’s audiences in front of the royal palace may have been a traditional form inherited from earlier times.

Furthermore, through an examination of the wooden structures depicted in the bas-reliefs of the Bayon Temple, Groslier speculated on the various components of the former royal palace. These included the guard corridors, courtyards, corridors for palace officials, and the palace harem. Visitors would pass through these areas to be granted an audience with the king or prince, signifying their socially prominent status [6]. In other words, it is conceivable that the royal palace comprised both a public audience space, such as the Elephant Terrace, and a more intimate and private audience space with the king.

The Book of Sui also records that the Great Hall had a front courtyard and gate, with a deployment of one thousand guards, suggesting the presence of complex facilities in the vicinity. As mentioned earlier, on the eastern side of M.90, front area of the royal city, composite structures with distinct features from the temples centered around M.138 and M.75 are identified. The area from M.90 to the outer wall on the eastern side of M.138 spans approximately 500 m, resembling the scale of royal palaces during the Angkor period. While caution is necessary when directly comparing the powerful centralization of the Angkor period with the early state formation period, it should not be denied that a complex royal palace existed in Īśānapura, featuring the king’s residential buildings and administrative facilities. It is reasonable to speculate that such facilities were integrated around the Great Hall, forming an interconnected complex. However, unlike the palace in Angkor Thom, which featured a large terrace in its forefront, in Īśānapura, terrace-like structures were positioned behind various facilities, suggesting a slightly different function for these terrace-like structures.

Viewing the overall layout of the city of Īśānapura, M.90 is likely situated at the farthest point, serving as a highlight along the visitors’ route. In other words, visitors who arrived on the western bank of the Sen River, which flows on the east side of the city, would have first paid their respects at the temples, a symbol of the state prestige, after traveling over a 2 km-long causeway (Figure 2). Subsequently, they would proceed to the western royal city area, passing through the facilities of the royal city entrance complex arranged along the city’s axis, before reaching the Great Hall, the final destination for an audience with the king. It is speculated that such an extensive sequence was designed to enhance the king’s authority through a grand urban plan, with the Great Hall at M.90 serving as the ultimate point where the king would be seated, symbolizing his power and majesty.

6. Conclusions and Future Work

The structures unearthed from M.90 have been confirmed to be unique not only within the confines of this archaeological site but also across ancient Khmer architecture in general. Situated symbolically at the approximate center of the square city, it consisted of a terrace measuring 2.1 m in height and 47 m north–south by 15 m east–west, along with an accompanying chamber to the south, clearly differing from typical temple architecture. The terrace would be completed in the early 7th century, with modifications adding brick structures to its surface in the first half of the same century, suggesting the possibility of wooden buildings being erected on the terrace.

The Book of Sui, a Chinese historical record compiled in the early 7th century, contains detailed descriptions of Īśānapura, which has been identified with this complex, and includes reference to a “Great Hall” where the king carried out state affairs. Based on its symbolic location, scale, and inferred construction and modification dates, the large terrace excavated from M.90 is considered the only structure that can be reasonably identified as this “Great Hall.” This identification provides a new avenue for considering how royal governance was conducted in the past, the architectural composition of administrative buildings, and the layout of political facilities and administrative structures within the city.

The audience hall is presumed to have been closely related to the king’s private palace quarters, forming an integrated complex in both spatial and functional terms. In later Angkorian urban layouts, royal palaces were typically located either to the north or south of the state temple, which faced east, placing the two in a parallel relationship. In contrast, Īśānapura is the only known case where the palace was placed to the west, that is, behind the temple mountain, within the city layout. This unique configuration likely reflects the need to foreground the temple as a symbol of royal authority and implies that the urban planning of the time was underpinned by a theocratic system of governance. This layout may also be interpreted as an earlier phase of governance preceding the emergence of the more integrated concept of the devarāja (god-king), in which the king and deity became ideologically unified and temples and palaces were positioned in parallel.

Furthermore, considering that known Angkorian royal palaces were complex facilities, it is plausible that administrative and residential structures surrounded M.90. Descriptions in the Book of Sui of numerous guards and associated facilities provide supporting evidence. Based on LiDAR-based topographic surveys and surface investigations, a group of complex remains has been identified east of M.90, suggesting the presence of palace-related structures. Located at the innermost point of this compound, M.90 may have been the most important building where the king personally conducted state affairs. The spatial arrangement of such administrative buildings within the royal palace offers new insights into the king’s role in domestic and foreign governance and into the changing administrative systems from the pre-Angkor to the Angkor period. To further substantiate the identification of M.90 as the Great Hall and to investigate the surrounding facilities, additional archaeological research is essential. The partial excavations conducted thus far have not uncovered the luxurious decorations described in the Book of Sui, nor have they revealed any staircase providing access to the terrace where the king would have sat. Broader surface excavation on the terrace mound is required to elucidate the presence and structure of the wooden superstructure. Moreover, archaeological investigations targeting the surroundings of M.90 are necessary to verify the presence of a complex palace.

Identifying the Great Hall described in the Book of Sui and reconstructing the complex royal palace not only clarifies the core structure of Īśānapura’s urban plan but also contributes to the broader study of the evolution of ancient Khmer cities and royal palaces. Such investigations are only in their initial stages, and continued research will be crucial going forward.