1The so-called ‘Four Thousand Islands’ (Si Phan Don) in Champasak province in South Laos is an area where literally thousands of islands and islets are scattered over a 40 km long stretch of the Mekong River which is divided here into a multitude of larger and smaller branches (Fig. 1). These branches reach eventually a hard Cretaceous greywacke formation that forms a geographical obstacle in the landscape, causing the water to plunge over a ten km wide east – west string of several narrow natural cascade-like channels and dangerous rapids (Fig. 2). It may be defined more accurately as a cataract waterfall with an irregular steep surfaced gradient and a vertical drop of approximately twenty meters. At present, the cataract waterfall is delineating a section of the border between Laos and Cambodia . Among the five southernmost islands Don Khon is the largest and obviously the only one featuring archaeological remains related to the Khmer period (Fig. 3). Situated within the compound of the Buddhist Wat Ban Khon Tai, southwest of Ban Khon Tai and about one km northeast of Somphamit/Li Phi waterfall, scanty vestiges of a Khmer temple are still visible (Fig. 4). If a visitor is unfamiliar with the location, they are very easy to miss. This might explain why Santoni (Santoni 2008: 103) does not mention the site in her survey of Khmer temples in the villages on the banks of the Mekong in Champasak province, whereas some years later in a short note Lorrillard (Lorrillard 2012: 194) refers to it as Wat Don Khone [sic] and charts it on a map of ancient Khmer sites.

2At first glance, it would appear that the site reveals just another example of the numerous small and dilapidated temples which are dotting the landscape of the Champasak plain, west of the Mekong. In this case, however, a more detailed analysis might reveal an archaeological site of particular interest. The remains of the temple consist of a brick platform that carries a Lao stūpa-chapel[1] and the foundations of a destroyed tower (prasat B). Several sandstone pedestals are scattered over the monastic compound. Together with a liṅga, a praṇāla and a decorative as well as a functional lintel with two structural reveals they prove the existence of a Khmer temple. Although the major part of the ancient temple is lost, the arrangement of the remaining traces of prasat B as well as the dimensions of the platform allow to draw some tentative conclusions on the original temple layout.

3The brick platform with the stūpa-chapel and two congregation halls (sim) dominate the central area of the Buddhist monastery. The stūpa-chapel and the older of the two sim as its most important buildings use the platform as their foundation, while the modern congregation hall was later added to the northeast (Fig. 5, b). The living quarters of the monks (kuti) are located beyond the platform in the southeast part of the compound.

4The northeast side of the platform with the exception of the east corner vanishes under the older sim and a twentieth century concrete platform that features a small Lao that and some sculptures illustrating major events in the life of the Lord Buddha (Fig. 6). The platform’s east corner together with the northwest extension is certainly a later addition (Fig. 7). The same applies also to its southeast edge which doesn’t represent the original limit of the platform because some of the laterite curbstones, which are mixed with bricks, indicate that they are spoils from other constructions (Fig. 8). Thus, the exact original perimeter of the Khmer temple platform is no longer visible due to the destructive forces of nature and several modifications in the course of time.

5Another important aspect is the spatial position of prasat B in relation to the platform edge. The spacing of more than seven m between the tower foundation and the southeast edge of the current platform exceeds by far the usual standards in Khmer architecture even for larger constructions. It is rather unlikely that the Khmer builders had deviated from the architectural norm in such an unusual way. On the southwest side an unevenness in the ground might be interpreted as the indistinct contour of the platform, though in my opinion it is nothing else than rubble from the destroyed temple buildings concealing the true course of the edge. And finally, on the northwest side the waters of a branch of the Mekong eroded the platform considerably in the course of time. There, a new rim of re-used laterite blocks, which does not represent the original edge of the platform, was constructed in order to protect the foundation of the two Buddhist sim from further erosion (Fig. 9, a).

6With a height of approximately 0.55 m the platform is very low for Khmer building standards. It still has an even surface and obviously retained its original occupation layer, including the floor of prasat B which has partly been preserved. In contrast to the towers, it has largely escaped the fate as a source for building material. Everywhere on the platform traces of brick flooring in situ are visible, continuing under the thin grass layer. Several meters northwest of prasat B a small heap of bricks is intermingled in the roots of a fig tree. Possibly, these loose bricks belonged to another destroyed tower.

7In its present state of conservation the platform measures roughly 41 m northeast to southwest and 32 m northwest to southeast, however, it has to be admitted that after the last expansion it must have covered a surface of at least 41 x 44 m. When this final structural alteration work was carried out is difficult to say, but it possibly might be dated to the time when the Lao settled the area and reused the former Khmer temple.

8Among the buildings of the monastery is a typical Lao that which was constructed from laterite and brick with a whitewashed plaster coating. It rises in front of the southwest wall of the older sim and marks the approximate center of the brick platform. A stucco inscription that is now completely lost disclosed a restoration work in 1931[2] but one may assume that its initial construction dates much further back in time. As mentioned above on closer examination the alleged that is in fact a stūpa-chapel since it has a small chamber accessible from the northeast through a low doorway (Fig. 10). Viewed from the outside the chamber of the stūpa-chapel is designed as a three-tiered base with a height of 2.40 m supporting a śikhara-shaped top. The outer dimensions of the lower tier measure 5.53 m on the southeast and 6.60 m on the southwest side.

9However, the stūpa-chapel disguises with reasonable certainty the foundations of a former Khmer prasat. Obviously, the Lao re-used the material of this prasat as well as the other dilapidated Khmer towers and constructed from the spoils a completely new and presumably slightly smaller building. Bricks and laterite blocks were jumbled in the walls (Fig. 11) which are neither mirroring the masonry nor the floor plan of the original Khmer building. With certainty we can assert that no parts of the walls are in situ at all, and also at least one of the door jambs of the northeast entrance seems to be not in its original position. The outer dimensions of the second tier measure 5.60 x 5.05 m which would be even undersized for the central prasat of this small temple complex. The true dimensions of this former prasat should at least slightly exceed the ground plan of prasat B. The building is surrounded by a 0.20 m wide concrete rim or reinforcement measuring 8.15 x 6.60 m. This rim might adumbrate the external foundations of the former Khmer tower[3]. If this is the case the dimensions of approximately 9 x 9 m for prasat A match the size of the main and usually larger central tower in a Khmer temple. Today the chamber serves as a repository for the cremated remains of the resident monks that are deposited in simple urns. Against the inner southwest wall of the chamber a sandstone slab is placed, which originally might have served as a sīmā-stone. It is adorned with a relief of a seated image of the Lord Buddha, executed in a rough rural Laotian style (Fig. 12).

10Nearly completely buried in the rubble of the cramped confines between the stūpa-chapel and the southwest wall of the older sim is a second sandstone slab which is placed on the left side of the doorway (Fig. 13). In contrast to the sandstone slab inside the chamber its shape is carefully executed and shows slightly rounded shoulders, however, without any visible decoration, except for a short groin on its top middle. There is no doubt that it is either an inscription stele or a sīmā stone and possibly even a combination of both. The shape and design of its visible upper part is reminiscent of sīmā stones of Mon/Dvaravati on the one hand and post-Angkor origin on the other hand. According to the classification of Murphy (Murphy 2010: 345 and table 6.1) it seems to be identical with the most common slab-types 1 or 8 sīmā typical for Northeast Thailand and Central Laos, which are chronologically ranging from the ninth to the eleventh centuries. During this period the south of Laos had been a Khmer dominion for at least three centuries and a Mon influence seems to be rather unlikely and is at best difficult to explain. Viewed from this perspective, a correlation with the late Angkor- or even post-Angkor period must be considered because the slab’s upper part is matching with similar sīmā stones in Angkor Thom e.g., at west Prasat Top that dates to the thirteenth or fourteenth centuries. This would also mean that in all probability the temple was already converted into a Buddhist monastery or hermitage by the Khmer in the late twelfth or beginning of the thirteenth centuries, long before the arrival of the Lao[4].

11The sandstone doorway of prasat A, consisting of two jambs, a broken functional, and a more or less intact decorative lintel, is probably also not longer in situ but was re-erected on the modified ground plan incorporating the original material. The clear width of only 0.69 m between the jambs would be unusual narrow particularly for the entrance of the main tower. This assumption is further supported by the fact that the decorative lintel is projecting too far beyond the side edges of the jambs. Therefore, it can be assumed that at least one of the jambs was moved into a new position. The decorated lintel, which is now resting on the broken functional lintel and was bricked in, had been originally supported by two columns that are no longer preserved. Due to a massive infill of rubble within the chamber the doorway appears to be very low.

12The only visible extant work of art that allows a stylistic and chronological classification is a decorative lintel above the doorway of the stūpa-chapel (Fig. 14). The lintel with a length of 1.35 m and a height of 0.42 m is partly weathered but still in an overall good condition. Two crooked lateral floral garlands ending in curled foliage are forming each an inverted U and divide the relief into two registers. The central motif depicts a deity within a pointed trefoil arch, sitting in a kneeling position on a three-headed elephant (Fig. 15). The god’s upper torso and head are badly damaged, however, the lower bulged headdress and the earlobes or earrings are still faintly discernible. His left arm is stretched while the right arm is bent upwards, resting with its elbow on the right knee. The left hand touches the left knee whereas the right hand seems to hold an object. The god wears a typical Khmer sampot with regular plaits. The representation of the three-headed elephant with four (!) column-like forelegs is simple and stiff. The trunks of the lateral elephant heads are holding the garlands, while the middle one is pulling out a plant stem. The lower register shows curled foliage and on both sides of the elephant a small, pointed trefoil arch enclosing a fleuron (Fig. 15, b). The fleuron recurs once more on each side though barely discernible and without a trefoil arch. In the upper register a triśūla (Fig. 15, a) is alternating with a trefoil arch cum fleuron, the former depicted three times the latter twice on each side of the deity. With the exception of the upper side, where perhaps a row of deities or celestial beings completed the relief, an undecorated rim frames the whole scene. Undoubtedly, the lintel depicts the god Indra, who is the guardian deity of the east, riding his mount airāvata. Although the relief shows in general the conventional iconography of a Khmer lintel facing east, it is remarkable to find the triśūla in this context, and this even if the temple was certainly dedicated to Śiva (see below).

13From an artistic point of view the carving is not very demanding, but the sculptor lent at least some dynamic expression to the depiction of the god Indra as well as the foliage. The lintel belongs to the many similar copies that were executed in local workshops by local artists on the artistic periphery of the Angkorian empire as Polkinghorne (Polkinghorne 2007: 230) proposes. According to the schema of Jacques (Jacques 1990: 177–178) and the remarks of Stern (Stern 1934) regarding Khmer lintels, it can tentatively be dated to the tenth century, an assumption that is also supported by the occurrence of the fleuron motif which was widely used by the artists in this period (Polkinghorne 2007: 226). Therefore, it is entirely reasonable to assign the decorative lintel to a sort of localized Pre Rup style.

14Another architectural element of the former temple was part of a praṇāla that once conveyed holy water from the main cella to the outside. It was re-used as a second ‘lintel’ spanning the gap between the two facing walls of the middle tier which are bordering the doorway (Fig. 14; Fig. 16). Although now void of its former function the Lao obviously regard this religious device as much adorable as did the Khmer and placed it in this prominent position, although, detached from its original architectural context.

15The laterite foundation of tower prasat B is still clearly visible (Fig. 17). Its outer dimensions are 7.10 x 6.67 m, while the inner chamber measures 4.00 x 3.87 m northeast/southwest and northwest/southeast, respectively. On the average the walls had a thickness of 1.50 m and were probably constructed from brick and partly laterite. On the northeast side two stones of the laterite masonry show a slightly raised broad edge that clearly marks the transition from floor to inner wall (Fig. 17, d; Fig. 18). Furthermore, according to Dumarçay (Dumarçay 2005: 65; Fig. 81, e) the edge could also be explained as a brace in order to increase the consistency of the horizontal laid courses of the masonry.

16Literally nothing remains of the building above ground level except for a course of laterite ashlar at its inner east corner as well as a cylinder and a pedestal made from sandstone (Fig. 18; Fig. 19). The pedestal (Fig. 17, a) has an edge length of 0.84 m at its base and a height of 0.45 m. Since the upper part is eroded and damaged it is impossible to say whether it held a statue or a liṅga, but its original height must have been approximately 0.50 m. Its sober design shows a central banded register bordered by two larger concave bands and can be classified as a pedestal style A1 according to the typology of Klassen – Kerwin (Klassen – Kerwin 2017). It rests on a sandstone slab that measures 1.15 x 1.17 m with a surrounding rim. The rim has a width between 0.165 m and 0.18 m and borders a shallow indentation that is fitting in order to mortise the pedestal. This plinth possibly still covers a foundation shaft which must be relatively short due to the low brick platform and the underlying hard bedrock. The carved square pedestal had been disturbed but is more or less in situ[5]. Beside the pedestal stands the cylinder (Fig. 17, b) with a height of 0.45 m and a diameter of 0.45 m. The top is decorated with an eight-petalled lotus flower motif. It certainly functioned as a roof finial which belonged either to prasat B or one of the other towers and is the only evidence for the lost upper structure of the sanctuaries.

17Noteworthy are two laterite blocks in the cella floor each showing a semi-circular notch close to the south and east corner of the former chamber (Fig. 17, c). They are not arranged at random but obviously mark the spots of, in effect, circular holes intended for the insertion of posts. Although their two counterparts on the opposite side of the chamber are no longer traceable, they could certainly only have served as a slot for two of the four wooden columns that originally supported a dais spanning the pedestal with its statue or liṅga.

18While the prasat foundation was built from laterite, finials and probably door frames were executed in sandstone. At least part of the cella floor was possibly covered with a sandstone pavement that is now missing. The bulk of the material must have consisted of baked bricks for the walls as well as bricks with additional sandstone elements for the roof. At any rate, the sandstone finial disproves a roof construction of light material. Since there is not any remaining brick wall layer in situ, the complete material had probably been re-used for constructions in the monastery or nearby village.

19On the site of the monastery seven more or less intact pedestals are extant. As we have seen only the one in the former cella of prasat B is still in situ, one seems to be carelessly discarded a few meters northeast of the modern sim (Fig. 20), three are stacked in a shed supporting a liṅga (Fig. 21), and two are placed at the east corner of the platform[6]. While six are belonging to the style A1 only one represents style B (Klassen – Kerwin 2017), which is characterized by a lack of the central register (Fig. 20). For the moment this classification is not yet appropriate for an accurate dating, but the pedestals all belong quite certainly to the Angkor period[7]. The number of seven pedestals further supports the consideration that there were more towers than the remains of prasat B might suggest. Due to the shape of their sockets at least two pedestals seem to have been designed for statues but so far obviously no image has been recovered. Solely on the southwest side of the stūpa-chapel, under the roof of a small shelter, the above-mentioned sandstone liṅga is placed (Fig. 22; Fig. 23, d). It is in a very good state of preservation and follows the typical classical Khmer design, consisting of cube, octagonal prism and cylinder. It has a height of 0.89 m and at its base a width of 0.29 m. Today, the liṅga rests on three superimposed pedestals (Fig. 20), which are arranged according to their size on a three-tiered base made from concrete or stucco-covered masonry. The uppermost pedestal has in its upper rim a notch for mounting a slab with a somasutra. The lowermost is the largest one and could originally have held the liṅga. Due to the size the liṅga doesn’t fit in the socket of the pedestal in prasat B, and its positioning as well as the quality of execution might indicate that it belonged to the cella of the main tower within the temple. The liṅga clearly reveals that the temple was dedicated to the god Śiva. The symbol of the triśūla on the decorative lintel further supports this identification.

20Although the physical remains of the temple complex are very modest, they nevertheless provide some indications which may elucidate its original layout. The longitudinal axis of the older ordination hall is very accurate on a line with the conjectured ancient Khmer processional roadway leading to the stūpa-chapel (Fig. 5, f, angle measurement α)[8]. From this observation we can argue that also the initial construction of the major buildings of the Lao Buddhist monastery – the older ordination hall and the stūpa-chapel – was based on the ancient Khmer temple orientation and layout. All the above-mentioned available data are plotted in the ground plan of the central part of Wat Ban Khon Tai and show the extant Angkor as well as the post-Angkor constructions (Fig. 23).

21A single tower prasat B can be ruled out because it would have been in a position on the platform, which is extremely off-center, virtually impossible for Khmer architectural standards as well as esthetics. The alternative of a layout with two rows of three towers each like Preah Koh of the Roluos Group must be rejected since the brick platform does not reveal any foundations of towers at the expected places. Unfortunately, towers C and D are completely lost, while the foundation of tower E is possibly still partly hidden beneath the above-mentioned concrete platform[9].

22An aerial view of Ban Khon Tai and the Buddhist monastery discloses an axis (Fig. 5, g) leading from the large ordination hall (Fig. 5, b) through the main monastery gate (Fig. 5, c) straight into the western area of the village. At first sight this axis seems to represent the ancient processional roadway belonging to the Khmer temple[10]. However, this alignment can be better explained with a redesign of the complex by the Lao probably in the more recent past because its orientation 45° to the northeast is incompatible with the different orientation of stūpa-chapel A and prasat B. The latter are oriented with 58° to east-northeast, after all a deviation of 13° which cannot be explained by flaws during the construction process and was planned purposely. It has to be mentioned that this figure differs slightly from the usual range of alignments of Khmer sanctuaries surveyed in East and Northeast Thailand, where most are oriented to within 25° of true east (Mollerup 2012: 155–156)[11]. Moreover, if we extend an imaginary straight line from the stūpa-chapel with an angle of 58° to east-northeast it intersects a part of the village east of the monastery gate, where, interestingly, a cluster of modern buildings is similarly aligned (Fig. 5, e). It remains to be seen whether or not this is a mere coincidence but in my opinion, it indicates that the modern buildings in the southwest part of Ban Khon Tai possibly occupy the area of a former Khmer settlement to which belonged the temple complex (Fig. 5, f, angle mesurement α). The spatial arrangement of the modern buildings in Ban Khon Tai might reflect the changes of the village layout through the ages and mirrors phases of destruction and rebuilding. Particularly, the northwest part of the village (Fig. 5, d) and the monastery which line the riverbank are prone to seasonal flooding. This resulted in the repeated deterioration of the houses built from light material; even the laterite and brick fabric of the temple suffered severe damage, up to the total loss of some structures in the course of time. This is the reason why the riverside area of the monastery is almost void of buildings and the brick platform in places completely eroded. The monastery compound is unusual large for the modest number of monks and for the most part its present perimeter could be identical with the former larger Khmer temple precincts. Aerial photographs do not disclose any further details of the ancient temple site like enclosures or gopura. The typical Khmer baray, which serves as a symbol for the cosmic ocean, is missing here because the location of the temple on an island bordered by branches of the mighty Mekong made a sacred water tank redundant.

23The question remains as to how the layout of the original Khmer temple was planned. It can be inferred that the temple featured multiple towers. The hard facts for this assumption are:

- the size of the platform is unusually large for just a single tower B,

- the foundations of tower prasat B are not placed in the middle of the platform but in the south corner,

- the stūpa-chapel A was most likely erected on the foundations of a predecessor building which is contemporary with prasat B and the platform because its walls are parallel to the foundation lines of prasat B as well as the southeast and northwest edge of the platform and thus follows the initial orientation conceived by the Khmer builders,

- the main processional causeway of the monastery runs in a straight line to stūpa-chapel A which implies that also for the Lao this spot and not prasat B was the focus of worship in the complex. In general, the major shrines were re-used for the cult of the new religion[12],

- seven extant pedestals bear witness to the existence of more than one tower.

24Moreover, is it possible to argue that the temple was even built as a quincunx of towers? Although the position of stūpa-chapel A and prasat B would allow for a reconstruction of either a double row of three towers on a rectangular[13] or a quincunx on a square platform the latter seems to be more plausible inasmuch as there are definitely no remains of foundations adjacent to the southeast and southwest side of the stūpa-chapel. Thus, the spatial arrangement of stūpa-chapel A and prasat B in relation to each other as well as to the platform edges is an indication that the builders had probably a quincunx in mind.

25The edges of the present platform (Fig. 23, g) obviously encase the ones of the former Khmer temple. Understandably, the line marking the edge of the Khmer temple (Fig. 23, f) is conjectural but should correspond approximately with its original course. As a matter of fact, the quincunx corner towers of the state temples at Angkor were usually built with a distance of roughly two meters from their platform edges. With reference to this figure line f delineates cogently the conjectural edge of the ancient Khmer temple platform and, interestingly, coincides with the northeast end of the old sim. The old sim seems to have been built on and, what is important, exploits exactly the complete space of an empty lot between the stūpa-chapel/prasat A and the northeast edge of the Khmer platform making use of its foundation. This would also explain the building’s unusual square-shaped ground plan. Thus, the original platform could have covered a square-shaped area of approximately 32 x 32 m.

26The assessment of the available data justifies with a certain probability the cautious conclusion that the complex was built or at least planned as a quincunx of towers. Should this prove to be correct it will be the only one discovered in Laos so far.

27The quincunx as an architectural model has its origin in India and is ultimately based on the specific symmetrical pañcāyatana (see below) which is always rotational symmetric with regard to the rotation axis of the central tower. Furthermore, if the five towers are more or less identical in construction as well as design and share the same platform this configuration may be termed as a true quincunx. Although its layout is certainly based on formal considerations other aspects also play a decisive role. The distribution of the five towers in a quincunx is a slight variation of the visualization of mythical Mount Meru, the central cosmic axis with its four minor peaks or buttress mountains accentuating the cardinal directions. The inserted two graphic symbols illustrate the difference between a true quincunx ⚄ and the layout of mythical Mount Meru ✜. With an arrangement of the towers in a quincunx, the master builders achieved a well-balanced and harmonious appearance which is esthetically pleasing. However, this is with the compromise that the ‘buttress mountain’ towers are not any more accentuating the cardinal but the sub-cardinal points. Irrespective of this reservation the quincunx is still a significant symbol of the cosmic model as Mabbett (Mabbett 1983: 74) noted with regard to the design of the Viṣṇu temples at Deogarh in India and Angkor Wat in Cambodia. The former is one of the earliest surviving true quincunxes in Hindu architecture dating to the sixth century, the latter the pink of their architectural perfection. The moats and water tanks built in connection with the towers symbolize the mythical oceans surrounding Mount Meru.

28True quincunxes represent only a fraction of the wealth of the Khmer architectural heritage and their factual number is probably limited to less than a dozen buildings, although Chihara (Chihara 1996: 159) surprisingly claims that there are many examples of this type of layout. The explanation for this bias lies in the fact that due to the prodigiousness and particular awareness level of temples like Angkor Wat one receives a false impression of the real prevalence of the quincunx. It materialized first and foremost in the monumental quincunxes at Angkor that were all commissioned as state temples by Khmer kings:

- Phnom Bakeng,

- East Mebon[14],

- Pre Rup,

- Ta Keo,

- Angkor Wat.

29Although any other quincunx didn’t exist of comparable size in ancient Kambudjadesa it seems to be obvious that their architectural models were designed much earlier at Sambor Prei Kuk after the consolidation of Chenla under king Ísanavarman I in the first half of the seventh century. Sambor Prei Kuk became something of an ‘experimental laboratory’ for early Khmer architecture which influenced the development of pre-Angkor and Angkor temple layout, construction techniques and design (Gabel 2022: 124). Nevertheless, up to the present day only a few other modest-sized quincunxes are documented beyond Angkor, and most are known by archaeologists only. The following list is certainly not exhaustive on the one hand, and on the other hand it probably contains already almost all examples which had ever been constructed:

- Prasat Sambor and M39 (Ichita 2021: 56, Fig. 2) at Sambor Prei Kuk,

- Prasat Chaom Sram in Preah Vihear Province (Ichita 2021: 71 , Fig. 20),

- Prasat Sikkhoraphum in Northeast Thailand.

30The existence of quincunxes both at Sikhoraphum, approximately 150 km north of Angkor, and in Preah Vihear province, is a bit obscure and worth the effort of being further investigated[15]. At present it is impossible to find a satisfying answer to the question why they had been built precisely at these locations. We can only speculate whether the sites had a certain religious, economic and strategic significance or were for an unknown reason recognized by the award of granting the construction of a quincunx. Possibly, a local Khmer chieftain simply challenged the authority of the king in Angkor by initiating a temple with a design in effect reserved for the royal seat of power. This is true also for the little-known quincunx of Prasat Chaom Sram that was built on a terrace comparable in size with the one of Prasat Sikhoraphum. In contrast to the latter temples, it has a substructure consisting of at least two tiers as well as several additional buildings within the inner enclosure. Its overall architecture is reminiscent of the East Mebon and albeit it was apparently not a state temple, it possibly marked a site of certain importance (Ichita 2021: 71, Fig. 20). Be that as it may, the construction of a quincunx beyond Angkor was obviously an extremely rare occurrence and for the surrounding area had an outstanding religious significance. Probably, its construction had to be authorized or sanctioned by the highest authorities. Due to the fact that the quincunx at Sikhoraphum is extant and by and large retains its original appearance it can serve as a model in order to visualize the hypothetical layout of former Prasat Ban Khon Tai (Fig. 24)[16]. It should be mentioned that the quincunx also appears in Khmer art which can be exemplified by a group of liṅga forming a true quincunx arranged within a yoni carved from the riverbed at Kbal Spean in the Kulen Mountains (Fig. 25).

31The term pañcāyatana, which literally means ‘five shrine’ (Huntington – Huntington 1985: 207), appears first in a corpus of Indian ritual literature where the pañcāyatana pūjā is mentioned in connection with the worship of the five highest deities, and these rituals influenced to a certain extent the layout of temple architecture (Reitz 1999). In India the earliest pañcāyatana complexes have been identified at the latest in the fifth century for Buddhist and the sixth century for Hindu architecture (Reitz 1999: 228). The Khmer seem to have embraced the pañcāyatana idea not long after in the early pre-Angkor period, while the Mount Meru cosmogony and the quincunx were adopted no later than the beginning of the seventh century[17]. The conjunction of both concepts resulted in an appropriate architectural solution that can be termed pañcāyatana-quincunx[18].

32Prasat Sambor with its slightly raised tower on a central terrace might be regarded not only as a kind of prototype for the temple-mountains in Khmer architecture (Ichita 2021: 79), but also here, for the first time, a pañcāyatana layout was implemented[19]. The main tower N1 is surrounded by the satellite towers N7 to N10 which are placed at the corners of the inner enclosure. Although N7 has a different layout than the other towers and they do not share the same platform with the central tower N1, Datta – Benyon (Datta – Benyon 2016: 164) inaccurately term the arrangement a quincunx and compare it with the Viṣṇu temple at Deogarh[20]. The remains of another four small corner shrines (N2 to N5) on the central platform are not contemporary with the main tower (N1), which dates from the seventh century (Ichita – Etsuo – Kojiro 2019: 1949; Fig. 11). They were added later and changed the design of the central enclosure which subsequently bore probably a close resemblance to a quincunx. This modification was not undertaken before the early tenth century and corroborates the presumption that the pañcāyatana-quincunx was not yet introduced in the pre-Angkor period. Perhaps, also the still unexplored small quincunx M39 at Sambor Prei Kuk is a construction of the Angkor period.

33The first real Khmer temple-mountain is Prasat Ak Yum and although it can be termed a pañcāyatana we can not define it yet as a true quincunx. The simple reason for this is that the central tower and the four corner towers are standing on different tiers of the pyramid. While Jacques (Jacques 1990: 43) inadvertently refers to it as a quincunx, Bruguier (Bruguier 1994) in his paper on Ak Yum deliberately avoids this term at all. If we exempt the ‘pseudo’ quincunx of Prasat Ak Yum from the later pañcāyatana-quincunxes we clearly see that the Khmer constructed the latter within a time period of roughly 250 years commencing with the temple-mountain of Phnom Bakaeng under king Yasovarman I around 900 CE. Thus, the theoretical concept of the pañcāyatana-quincunx was for the first time realized in stone. In my opinion the pañcāyatana metaphysics and its influence on Khmer religion have thus far hardly been given attention and deserve to be studied more intensely. The Khmer certainly integrated or modified some of the religious connotations which were inextricably attached to the imported architectural models in order to adapt them to their own spiritual beliefs. The symmetric pañcāyatana became indigenized in accordance with the worship of the most important Khmer deities and spirits (anak ta) allocating them to their appropriate shrines.

34Phnom Bakaeng and all subsequent state temples can be regarded as an architectural expression of power consolidation in the Angkor period because their realization required high economic strength and a strong leadership. The architecture of the quincunx was further elaborated in the course of time and finally culminated in the splendid and elegant central towers of Angkor Wat. Only during the eleventh century no state temple surmounted by a quincunx had been erected at Angkor.

35In the end, the mature quincunx in Khmer architecture is an accomplished amalgamation of the indigenous prehistoric terraced sanctuary in the shape of a stepped pyramid and the Indian concept of Mount Meru combined with the symmetrical pañcāyatana. Prehistoric stepped terraces were not only widespread in insular Southeast Asia but also a prevalent feature in the mainland, but subsequently they had been either reused as substructions for, e.g., Buddhist stūpa or even completely encased in younger buildings (Gabel 2022: 68 ff.). Commenting on city planning in the kingdoms of Southeast Asia Stuart-Fox – Reeve (Stuart-Fox – Reeve 2011: 105) remark ‘The location of cities, their orientation and their layout were not decided at the whim of kings: they required the combined talents of the most learned men at court – astrologers, geomancers and court Brahmins well-read in the religious texts of ancient India, notably the treatises known as śāstra on such diverse subjects as law, statecraft and architecture‘[21]. This applies particularly also to the temple as it is usually the very heart of the settlement. The Khmer builders seem to have observed the matter of the śāstra more strictly than the Indian master builders and finally created the temple-mountain surmounted by a quincunx, a genuine religious monument that, according to Jessup (Jessup 1997: 101), was not known in India[22]. It is not an exaggeration to state that the religious symbolism and architectural equilibrium of the Khmer quincunx surpasses even its equivalents on the Indian subcontinent.

36It is understandable that for the construction of the temple Don Khon was not chosen at random by the Khmer builders. The island constitutes the southernmost as well as the largest of the ‘Four Thousand Islands’ and is surrounded by roaring rapids. The natural spectacle must have attracted people from time immemorial and without doubt there had been ascetics meditating on the islands close to the cataract waterfall long before a temple was built. I agree with Lorrillard (Lorrillard 2012: 194) that the site of Wat Ban Khon Tai was possibly occupied in the pre-Angkor period as the area was part of Chenla but only careful excavations might verify such an early foundation. The island has sufficient arable land which provided the necessary surplus for the initial construction costs as well as the sustentation of the temple[23]. The natural and geographical setting was not only appropriate for a temple dedicated to Śiva but in the eyes of a learned Brahmin it must have been a location perfectly suited for a pañcāyatana-quincunx[24]. The visualization of a pañcāyatana-quincunx is based on the Indian concepts of cosmogony and mythology that try to explain the creation as well as the nature of the universe.

37The Mekong cataract waterfall is a unique geographical obstacle which causes the relatively calm waters of the river to flow over a short distance through several rapids in cataclysmic-like torrents. It is truly a magnificent spectacle displaying the forces of nature. The antagonism of both the life-giving and destructive waters are here particularly apparent and without doubt the Khmer were deeply inspired by the intriguing location. These natural conditions almost inevitably led to the construction of a temple dedicated to the god Śiva who epitomizes all of these forces.

38Embedded within this exceptional geographical setting, the main temple tower would have been a perfect rendering in stone of the Mount Meru concept: located on the northwest edge of Don Khon it symbolized the cosmic axis, while the roughly triangular-shaped island itself could have been conceived as the mythical continent Jambudvīpa. The temple did not need any artificial moat or a baray because it was standing on the banks of the Mekong which encloses the island and was considered as the Ganges River. Although the collation of the data seems to point in the direction of a pañcāyatana-quincunx on Don Khon it should be emphasized that in the end its existence can only be verified with reasonable certitude by an excavation[25].

39While the pañcāyatana-quincunx at Wat Ban Khon Tai might have lacked the grandeur of a pyramidal substructure it would have symbolized more than any other Khmer temple the ‘physical geography’ of mythical Mount Meru. But even if a quincunx cannot be proven one thing is certain, the complex was definitely planned as a temple constructed as an arrangement of more than two towers sharing the same platform which is hitherto unparalleled in Laos.

40Beginning with the consolidation of Lan Xang under king Fa Ngum the area of Si Phan Don became gradually part of the Lao border region[26]. Perhaps the former Hindu temple had already been converted into a (Mahayana) Buddhist sanctuary in the late twelfth or beginning of the thirteenth centuries, and the Lao could easily integrate its surviving fabric with their own architectural conceptions, serving now the ritual requirements of the Theravada Buddhist religion.

41For reading the manuscript I am grateful to Tonya Keane (Sligo) and Jonathan Gabel (Chiang Mai). Jonathan Gabel also managed to solve all technical problems and prepared the maps and plans. Sincere thanks to Jürgen Caspary (Wolfenbüttel) with whom I had inspiring discussions.

The remains of a small Khmer temple hidden within the compound of Wat Ban Khon Tai on Don Khon in South Laos were thoroughly investigated. Although these remains are in a poor state of preservation, the archaeological evaluation together with the special geographical situation at the Mekong cataract waterfall could indicate the former existence of a multi-towered temple or even a quincunx. The true quincunx is an exception in Khmer architecture and was essentially reserved for a small number of large monuments at Angkor. Beyond Angkor it occurs only at a few other sites of the Khmer empire but of more modest dimensions. The arrangement of five towers mirrors not only the Indian cosmogony focused on Mount Meru but also the pañcāyatana concept. Together with the stepped pyramid of indigenous origin the Khmer finally created an amalgam of Indian and local architectural perceptions culminating in the state temple-mountains of Angkor.