Introduction: Visualisation and Simulation
Over the past couple of decades, movement and perception have become major focuses of Classics.1 However, since they are both rooted in the body, they are embodied, as is experience in general (Merleau-Ponty 1945: 171, 235–237; Gibson 1979: 104–118; Stefanucci, Saxon & Whitaker 2021). This presents a significant challenge for archaeology, which relies on material evidence, as the transient nature of the body inevitably results in its loss. This is particularly important when the emphasis is placed on built space as a relational category consisting of landscape, architecture and living bodies that are constantly arranged and rearranged by movement and sensory perception (Lefebvre 1994; Schatzki 2015; Löw 2016). If we focus only on the fossilised aspects of such spaces, we miss the bigger picture and the process by which these spaces are socially produced.
Several methods have been developed in archaeology to address this problem. By locating and mapping certain kinds of action within dense archaeological records, we can gain an understanding of ancient smellscapes and soundscapes (Melini 2012; Nissin 2022). When it comes to ancient paintings, they can be analysed in their former context to understand the social construction of viewing (Ehrhardt 1991). If the focus is on the biomechanical reaction to certain architectural features, scholars may use their own bodies at ancient sites as a kind of scale to measure such impact (Hollinshead 2015: 22, 2019: 250–254).
Yet all of these approaches (including the one to be discussed here) are limited by incomplete preservation of material evidence. The material record excavated at a given site is rarely preserved or well enough understood for us to use our imagination and/or our own bodies alone to gain more than a mere glimpse of past social space. We therefore need methods to help us bridge this gap, such as visualisations or simulations.
Visualisations set up for the purpose of studying building history sometimes include small figures of ancient people to serve as a scale, as is the case with the bouleuterion in Ephesos (Bier 2011: maps 6, 7). Thus, visualisations can establish an initial connection between the ancient body depicted and the architecture by relating it to the proportions of the modern viewer’s body. Other commonly used forms of visualisation focus on movement or sight; for example, space syntax and viewshed analyses. These tools enable us to explore past accessibility and fields of vision within a well-preserved, defined area. Pompeian houses are a good example (Anderson 2023).
Since visualisations result in separate images, computer simulations can incorporate timespace. The agent-based modelling approach enables us to discuss human behaviour as inferred from a complex ground plan and a set of pre-defined rules, for example. We can therefore consider how initiates would move through a Greek sanctuary, for instance.2
All of these approaches are extremely useful in their own right. However, from the perspective of embodied movement and perception, they are limited by the seemingly objective distance they maintain between the body – whether ancient or modern – and architecture.
Game Engines
As our research questions evolve, so do the methods employed to bridge the gap between the body, complete with its motional and sensory apparatus, and the materiality of past spaces. Current developments in computer technology, particularly game engines, offer new ways to combine the advantages of 3D visualisations in general (Lanjouw 2016: 2–3) with real-time movement and embodied perception simulations. This enables us to examine ancient experiential situations from a praxeological, or practice-related, point of view.
Historical Game Studies is an emerging field within archaeology and classics (Rollinger 2020). When employed for scientific purposes, such studies tend to focus on two objectives in particular. Firstly, they analyse games with historical themes in terms of historical authenticity, how historical cultures are depicted, and the historical narratives created. Secondly, it involves the creation of new historical games or the gamification of a given situation, for example in classroom and university teaching, as well as for tourism. This second subfield of game studies is sometimes referred to as Serious Gaming, as it consciously uses game technology and/or game mechanics (digital and analogue) to motivate students, tourists, and the general public to engage with the historical material in question, raising public awareness of the importance of cultural heritage and its preservation (see, for example, the broad range of studies that follow: Kotarba-Morley et al. 2014; Zotti 2014; Oikarinen 2016; Varinlioglu & Muge Halici 2019; Buscemi et al. 2020; Politopoulos & Mol 2021; Heath 2022; Stocks & Birley 2022; Barbara et al. 2023; Winnerling 2025).
In contrast to the well-established fields of Historical Game Studies, a third field focusing on game technology as a tool used by researchers is currently emerging. Although analogue games are occasionally used in research as well (Arponen & Brinkmann 2025), many such studies – just like the ones mentioned above – make use of digital game engines. Therefore, these studies can also be categorised as Serious Gaming, despite the fact that game mechanisms such as contests or the implementation of motivational affordances to evoke behavioural outcomes (Hamari, Koivisto & Sarsa 2014; Oikarinen 2016: 28) are not usually employed (see Clinton & MacLaughlin 2019 for an exception). Instead, the ‘task’ may involve performing an action, such as conducting a sacrifice (adopting the perspective of a specific first-person character, such as a priest), and the motivation may be to achieve research objectives, such as gaining a praxeological understanding of the relationship between two sacrificial building units3 (adopting the perspective of the user/researcher). The product and process are therefore not usually games or gamified situations, but highly flexible, ‘playable’ or ‘playful’ research environments, which are usually governed by rules that approximate real-life conditions. Due to the complex mathematical algorithms already embedded in game engines, and their ability to be expanded with other software, it is possible to simulate physical properties such as gravity, sound propagation and visibility, which are shaped by different conditions (e.g. position, sound-absorbing textures and weather conditions), and to discuss the consequences of these properties for locomotion and sensory perception (Holter, Schäfer & Schwesinger 2020: 220).
Game engines can thus be deployed not only as a superior visualisation option for the purpose of testing pre-existing hypotheses, for example regarding the viewing axes between two entities (Frischer et al. 2017), but also as a means of facilitating the formulation of novel hypotheses. Even more importantly, game engines can be used to create concrete yet adaptable digital research environments that can be easily rearranged in terms of landscape features, architectural volumes and living bodies. These environments provide a foundation for discussing specific research questions in much greater depth (e.g. the social production of built spaces during festivals). Such digital environments enable us to focus on aspects that were previously only addressed briefly, if at all (e.g. where to herd sacrificial victims waiting to be slaughtered, or how to reach food preparation and banqueting areas when major road junctions were blocked by a large festival crowd).
The value of virtual environments in archaeological research has been recognised for some time. One of the earliest examples is a project by the OKAPI research group (Open Knowledge and Public Interest), who created a digital reconstruction of Çatalhöyük within the existing virtual environment of Second Life, to facilitate discussions about interior design and household furnishings in that digitally constructed space (Morgan 2009). As game engines such as Unity and Unreal have become more widely available, enabling researchers to create their own environments from scratch rather than renting digital spaces within existing games and adapting them, more and more scholars are taking advantage of these possibilities (Meister & Boss 2004; Ch’ng 2007; Anderson 2008; Rua & Alvito 2011; Frischer & Abdelaziz 2025; Muth et al. 2025). However, the number of projects using game engines not only to test reconstruction hypotheses, but also to create research environments in which to explore a variety of intertwined questions related to action and perception, is still comparatively small (see, for example: Hollinshead 2019: 246–248; Wescoat 2019; Holter et al. 2020).
Case Study: The Asklepieion of Kos During Festivals
With regard to the present paper’s focus on bodies, the virtually endless customisation options offered by game engines compared to inanimate 3D models help us consider bodies in two important ways.
Firstly, a digital 3D model can be populated with any number of human figures in any location, but the resulting image is inevitably static as these figures remain fixed in position. However, game engines open up the possibility of animating these figures, enabling them to move automatically along pre-defined paths. Therefore, not only the presence, but also the shift of bodies and groups of bodies can be taken into account when discussing the appropriation of space by the body.
Furthermore, while digital 3D models can only ‘beam’ the visual field to certain points of interest or move it along predefined paths like a camera, game technology enables the use of first-person avatars. An avatar is a figure in a game that is created and fully controlled by the designer and/or user. This provides unlimited possibilities for exploring the 3D environment within the pre-defined laws of physical reality (e.g. from the perspective of a priest) and walking in a natural flow, thus providing some kind of walking simulator (Morgan 2016: 10). Additionally, since game engines easily allow for the creation of virtual reality environments, the avatar can be used as an extension of the researcher’s own body, acting as a kind of prosthesis (Chrysanthi, Murrieta Flores & Papadopoulos 2012). This brings the researchers into more immediate contact with the virtual environment. It enables them to perceive the characteristics of paths and volumes in a size similar to real life through their own sensory apparatus.
To explore these options, the two authors of this paper initiated a DFG-funded collaboration4 between Freie Universität Berlin5 and the University of Applied Sciences Mannheim in 2024. The project, ‘Dynamic Spaces: The Sanctuary of Asklepios at Kos during Feast and Healing Ritual’, consciously draws on the merits of game engines for studying the formation of ancient built spaces. The Asklepieion, a 2nd century BC temple complex on the Greek island of Kos, was used as a case study. Its architecture was modelled using ArchiCAD software and embedded within the Koan landscape using Blender 3D software. It was then animated using the Unreal game engine. Users can then visually perceive this model in 360° using VR glasses (Meta Quest 3) and traverse it using a KAT Walk C 2 treadmill. This digital environment is used to compare and discuss the different types of built space that emerged during the aforementioned historical scenarios at the same site. The aim is to gain a better understanding of the decision-making processes behind the sanctuary’s design and how it interacted with humans during usage, depending on their respective roles and actions. However, the focus of this paper will be on a specific aspect of this broader field: the body-related logistics of the Megala Asklepieia, the major festival of Asklepios on Kos, as approached through movement and perception. By rearranging bodies in relation to the architecture and landscape, and using an avatar to explore paths within the model, we can discuss where different types of people and animals might have stood during cultic activities, and how they might have accessed their respective areas of interest.
The Building Stock
The Asklepieion of Kos was chosen as the paper’s (and the project’s) case study for two reasons. First, thanks to excavations, the site’s building stock is fairly well understood. Second, the site’s body stock during festivals can be grasped by drawing on a rich collection of textual sources.
Excavated between 1902 and 1907 under the direction of Rudolf Herzog (Herzog 1903b, 1903a, 1905; Schazmann & Herzog 1932: xxxi),6 the architecture of the Asklepieion was presented to the public in a monograph written by Paul Schazmann a couple of decades later (Schazmann & Herzog 1932). Some of the questions arising from this early research were addressed more recently by Wolfgang Ehrhardt (Ehrhardt 2014a, 2017, 2020).
Thanks to this research, we now have a fairly good understanding of the architecture of the Asklepieion from the first half of the 2nd century BC (Riethmüller 2005: 206–219; Interdonato 2013: 51–57; Rocco 2017: 339–341), modelled in the project (Figures 1, 2). At that time, the sanctuary consisted of three main terraces and an intermediate terrace on the northern slopes of Mount Dikaios in the north-eastern part of the island of Kos, around 3 km from the city of Kos.
Figure 1
Asklepieion of Kos, ground plan.
Figure 2
Asklepieion of Kos, bird’s eye view.
A pi-shaped Doric stoa framed the lowermost terrace to the north and the uppermost terrace to the south. The middle and intermediate terraces were bordered by retaining walls. The lowest level featured a wide, open courtyard containing water installations in the southern retaining wall. The main buildings on the middle terrace comprised a small Ionic antae temple at the western end and a small, two-part stoa (or fountain house?)7 to the south of this. At the centre of the terrace stood a large altar, surrounded by a high, pi-shaped wall. To the east of this was another building, a hall. The intermediate terrace is located between the middle and upper levels. It is shorter and narrower than the other terraces and has no known architecture (Herzog 1905: 5). Finally, the upper terrace featured a large Doric temple in the centre, probably situated within a cypress grove (IG XII 4, 1, 283–284; IG XII 4, 1, 71; Segre 1937: 194–197; Birge 1982: 462–465, nos. 313–315bis; Ehrhardt 2014a: 85–86).
The lowest terrace could be accessed via a broad propylon in the centre of the stoa, with a staircase in front of it. At the back of the upper stoa was a much smaller staircase that led up the hillside (Herzog 1905: 4; Ehrhardt 2014a: 77). The middle, intermediate and upper terraces were connected by two broad staircases. All four levels, including those bordered by stoai, featured large openings on the western and eastern edges (Herzog 1903b: 191, 1905: 6).8
The Body Stock
The ephemeral bodies connected to the Koan sanctuary left only a few traces, such as a worn slab from the paving between the columns of Temple B (Schazmann & Herzog 1932: 35). However, the combined written evidence from Kos (IG XII 4, 1–2) and other Asklepieia provides a fairly good picture of the body stock during festivals.
Generally speaking, a major Greek festival would consist of a procession, sacrifices and a banquet on one day, followed by competitions9 on subsequent days (Chaniotis 2011: 15). This was certainly the case at the Asklepios festival, which was held on Kos every four years. A lex sacra, a sacred cult regulation from the middle of the 3rd century BC,10 provides detailed information about those who took part in the procession (IG XII 4, 1, 286). As well as the priest and several cult attendants, it lists a group of animals destined for sacrifice: three cattle and two sheep for Asklepios, his father Apollo, his daughter Hygieia and his wife Epione. Furthermore, a series of texts from the same period indicate that a large number of envoys from other cities, theoriai (Rutherford 2013), were invited to attend the festival (IG XII 4, 1, 207–245). A decree from an Ionic city states that at least some of these ambassadors, theoroi, were given funds to contribute to the sacrifices on the festival day (IG XII 4, 1, 231).
However, these texts do not provide information about ordinary Koan residents. Yet, as the leges sacrae from other Asklepieia suggest, the authorities took great care to ensure the participation of the entire population (Eretria: IG XII 9, 194, 4th–3rd century BC; Lampsakos: CGRN 185, 2nd –1st century BC), although we cannot expect everyone to have played an active role in every part of the festival. Therefore, we can divide those participating in the festival into two groups: those who played a more versatile role in the procession, sacrifices and feasting, and those who played a more static – but no less important – role as observers of these activities. Moreover, individuals may have shifted between these groups.
Festival Logistics
A project like the one described here cannot be started from scratch; it must build on previous scholarship that considered where processions would have moved along and where sacrifices and banquets would have been held. The following section will critically evaluate and elaborate on this rich corpus of hypotheses, by discussing it against the background of the model.
As the model is still under construction, the human bodies are rendered in white, like the architecture and statuary. Furthermore, the number of bodies in the stock does not yet match the number of bodies described in ancient evidence (there are currently no sheep, for example).11 However, a sufficient number of bodies have been included to provide an indication of the direction in which the research is proceeding: 1,298 human bodies (with an average height of 1.70 metres for males and 1.60 metres for females), and three cows. It is not currently intended to explore the concept of embodiment with regard to the socially constructed individual’s body, although this would certainly be possible in order to challenge more normative views (Morgan 2022: 217). One might consider the question of how people with limited mobility would overcome the steep staircases to the sacrificial area on the middle terrace, for example, but this is more pertinent to another scenario (healing rituals). In the present paper, the discussion is limited to generic role models to explore the positioning of adults12 of higher status of both genders, identifiable by their elaborate garments (himation or peplos, respectively), and adults of lower status of both genders, identifiable by their modest garments (short or long chiton, respectively).
Procession (Video 1)
Video 1
Asklepios festival in Kos, procession sequence.
There is good evidence to suggest that the procession was assembled on the extremely spacious Koan Agora (Rocco & Livadiotti 2011) and then moved along a major thoroughfare, leaving the city through an western gate (Caliò 2017; Livadiotti & Rocco 2017). It then passed through the Koan plain, with its extensive necropolises,13 and finally reached the vicinity of the Asklepieion, at the point of a free-standing propylon approximately 150 metres to the north (Iacopi 1931: 480–481; Laurenzi 1931: 621; Bosnakis 2014: 23).
As Soi Agelidis has pointed out, we should picture festival-goers bidding farewell to the procession as it formed, and welcoming it again when it arrived at the sanctuary (Agelidis 2012: 95). Yet there is evidence that the Asklepieion was not an isolated architectural complex (Ehrhardt 2014a: 76–77), and we have no clear idea of the vegetation that grew around it in the past.14 Therefore, we cannot determine whether the area between the outer propylon and the propylon of the Asklepieion was suitable for accommodating a festival crowd (Figure 2). In contrast, the lower terrace featured a large open area. As the model shows, the courtyard and the adjacent halls could easily accommodate a large number of people alongside the procession. Moreover, the terrace’s outward-facing margins would have offered ideal vantage points from which to observe the procession as it completed its final metres to the altar area (Figure 3).
Figure 3
Observing the procession entry from the middle terrace (early afternoon).
Most individuals or small groups would have encountered few obstacles on their way to these areas. However, directing the movement of a large procession, especially one including large quadrupeds, was more complicated. The Asklepieion propylon was equipped with a free-standing staircase. This would have slowed the procession down. Nevertheless, the procession’s dignity would not have been severely affected, and if the (re)construction with three broad doorways measuring 1.20 – 1.60 m is correct,15 its members would still have been able to walk together in lines of two or three. However, such canalising doorways and staircases would have been less comfortable for animals. The steps, with an estimated rise of 17 cm and a horizontal extension of approximately 34 cm,16 were designed for humans, not animals (Hollinshead 2015: 19–24).
At this stage of the modelling process, the question arose as to whether there were any alternative routes with ramps that would be more comfortable for the cattle (Hollinshead 2015: 21). The lateral openings of the lower terrace and the ramp-like slopes of the hill bordering the sanctuary could have provided an appealing alternative to the staircases. However, two factors argue against this. Firstly, as the landscape modelling shows, these slopes must have been rather steep if the current situation resembles the ancient one at all (Figure 4). An incline of up to 50% when walking immediately next to the terraces would offer cattle no relief compared to the staircases, as an incline of more than 15% is unfavourable for them (Sporn 2015: 359 n. 74). Secondly, separating the animals from the procession would severely diminish its visual impact, as spectators inside the Asklepieion would not be able to see the animals as they approach. This contradicts the desire to incorporate the animals into the processional spectacle, which is evident in the selection of the finest specimens and the gilding of their horns (IG XII 4, 1, 286). Animal welfare was certainly not an issue here, as the sacrificial victims would only make this journey once. Therefore, the entire procession was modelled as proceeding up the stairs and through the propylon.17
Figure 4
Hillslope on the right-hand side of the Asklepeion (early afternoon).
Sacrifices (Video 2)
Video 2
Asklepios festival in Kos, sacrificial sequence.
The sacrifices followed the procession and would have occupied most of the day during the festival. This involved priests and other cult personnel moving around individually, as well as animals. Meanwhile, the majority of people formed a group of observers.
The only known altar of the Asklepieion was located on the middle terrace. Therefore, sacrifices to all the deities mentioned in the Koan lex sacra – Apollo, Asklepios, Hygieia and Epione – must have been performed there. Although the Koan altar table within the enclosure was spacious, we cannot expect all sacrificial victims to have been burned at once. Instead, we must imagine one animal after another being presented to the festival community before being slaughtered. The slaughtering and butchering would have been carried out by other members of the temple staff.
The priest stood inside the altar circuit, next to the altar table. However, this does not mean that all priestly activity was concentrated here. There is evidence that festive rituals may have included placing unburnt sacrificial portions on a cult table (trapeza) dedicated to a deity, a practice known as trapezomata (Gill 1974: 123–127; Ekroth 2008a: 95–98). Two such tables have been identified within the Asklepieion: one in the pronaos of the Ionic temple, opposite the altar on the middle terrace; and the other in the pronaos of the Doric temple, on the upper terrace (Kobusch 2022: 214–215). The Ionic temple most probably belonged to Asklepios (Kobusch 2022: 216–217)18 and possibly also to Hygieia and Epione (Laubscher 1961: 50), while the upper Doric temple may have been dedicated to Asklepios and possibly Apollo.19 As there were sacrificial victims for all of these deities, both cult tables must have been equipped with divine portions. Therefore, we must envisage the priest of Asklepios repeatedly moving between the altar and the Ionic temple, and between the altar and the Doric temple.
In previous versions of the model, the priest’s standing area was positioned behind the altar table, as indicated in earlier visualisations (Schazmann & Herzog 1932: Pl. 12). This meant that the priest faced the front of Temple B. However, this decision was not based on the physical remains, as the interior of the altar is not well enough preserved to indicate exactly where the prothysis, the steps leading up to the altar table, was situated (Schazmann & Herzog 1932: 26). However, when the avatar playing the role of the priest made his way from behind the table to reach the trapezai of both temples, it became clear that there would be little space to circumference the altar table within its enclosure. Therefore, in the model, the table was reoriented towards the inner wall of the altar, meaning that the priest now turns his back to Temple B (already suggested by Şahin 1972: 93–94). This is not a unique situation – most Greek altar tables seem to have been positioned this way with regard to their respective temples (Ohnesorg 2005: 221).
Given the large number of animals to be sacrificed – three cows and two sheep from the Koan citizens, plus an unspecified number of sacrificial animals from the foreign theoroi – we must also consider how to herd them throughout this long festival day. The middle terrace is narrower and more densely built than the lower terrace. Therefore, it would be inconvenient to keep all the animals next to the altar throughout the ceremony. A good solution would be to keep them on the spacious lower terrace near the retaining wall. There is a fountain there, so they would have access to water, and they would not obstruct the view of the interior of the altar for festival participants (see Figure 5). The difference in altitude actually obstructs the cow’s view of the altar and the proceedings taking place there. This, combined with the muffling of sound and smell,20 might have helped to keep the animals calm, as they would have been less affected by the singing accompanying the sacrifices, as well as the stench of bloodshed and burning offerings. In this scenario, then, the cattle would be guided up the staircase one by one when it was their turn.
Figure 5
Herding cows at the fountain house (early afternoon).
While it was advantageous to keep sacrificial victims physically and sensorially distant from the place of slaughter, the opposite is true for humans. The most prestigious places should be sought as close as possible to the cultic action, with the best view of the sacrifices. In the case of the Koan Asklepieion, only one area was consciously designed for viewing: a 3rd century BC exedra with an assumed seating bench next to the upper staircase (Freifrau von Thüngen 1994: 148–149). However, as the model shows (Figure 6), the view of the sacrificial action from it was rather limited due to the size of the altar circuit. Furthermore, although the exedra was exceptionally broad, seating was restricted to ten people sitting side by side. This would have been sufficient for only a small group, thus excluding the majority of festival participants. It is plausible that such a seating area would have been appreciated by elderly people, especially those of higher status, who were unable or unwilling to stand for several hours while sacrifices were being conducted.
Figure 6
Using the exedra as observation point (early afternoon).
There are no other obvious observation structures. However, the pronaos of the Ionic temple offered the best view of the area. Standing there, an individual would have been sheltered from the sun while enjoying a direct view of the altar circuit and the priest serving it (Figure 7). However, as mentioned above, we must consider the priest’s repeated entry for the trapezomata. Moreover, as the model shows, the presence of the trapeza substantially reduced the available space in this small room. With more than two or three people standing there, the space would have begun to feel cramped. Finally, the grilles between the columns created a partially translucent visual barrier (Kobusch 2022: 215–216), and the fact that the grilles opened inwards further reduced the available space. A larger number of observers inside the pronaos was not modelled for these reasons.
Figure 7
Observing the altar from the temple pronaos (early afternoon).
Another area that was certainly not designed as a viewing place, but which could nevertheless be used as such, would have been the upper staircase. Although its steps are too low and narrow for seating (17 cm high and 34 cm deep), they provided an elevated standing platform with an excellent view of the priest moving between the temple and the altar (Hollinshead 2015: 75). However, it is important to remember that a corridor had to be left for the priest to reach the upper temple. Nevertheless, the area was spacious enough to accommodate many of the most important festival attendees.
The intermediate terrace was the most convenient area for a relatively large group of people to observe the sacrificial action. Although rather short, it had no ancient buildings (Herzog 1905: 5) and was broad enough to accommodate the main figures of interest. It is even conceivable that this area was equipped with temporary wooden tribunes.21 Due to its elevation, the intermediate terrace could be accessed from both sloping ends22 and provided a good overview of everything happening in front of the altar circuit (Figure 8).
Figure 8
Observing sacrifices from the intermediate terrace (sunset).
As shown in the model, the most prestigious, albeit confined, areas around the altar would have been crowded with people of higher status. Everyone else – the majority of Koan residents, enslaved people and occasional visitors – would have had to make do with places further away, such as the edges of the upper and middle terraces, or wherever else they could catch a glimpse of the ritual proceedings.
Feasting (Video 3)
Video 3
Asklepios festival in Kos, feasting sequence.
The tightly packed central area of the Asklepieion has consequences for the subplot of cultic action. Not only were gaps or corridors necessary for the priest to reach the trapezai, but they were also needed for cult personnel to carry the remains of the slain animals away for processing in preparation for the feasting. Due to the buildings, the crowd and the ritual action, there was little room for cooking facilities on the middle terrace. A more plausible location would therefore be the lower terrace, which has a spacious courtyard. Few people would have stayed here to watch the sacrifices being made, since the view was largely obstructed by the staircase and retaining wall, not to mention the distance. Furthermore, the lateral exits of the sanctuary, as well as the steep slopes of the hill, which were unconvenient for cattle but manageable for humans, offered an opportunity to bypass the crowd.
In some sanctuaries, there was an ou phora law that prohibited the removal of sacrificial meat from the precinct (Goldstein 1978: 51–54, 322–355); everything had to be prepared and consumed on site. There is no evidence of this practice in Kos. Theoretically, therefore, we can envisage the portions being handed out to festival participants to eat at home. However, given the amount of meat to be processed and the number of people joining from abroad, this isn’t very probable (Ekroth 2008b: 280). It therefore seems more plausible to look for banqueting areas within the Asklepieion itself.
A couple of years ago, a hypothesis was proposed that identified the building next to the Ionic temple as a small banqueting hall with the typical layout of two rooms behind a transverse hall (Livadiotti 2013). However, more recent research by Ehrhardt suggests that this Hellenistic building was actually a small, two-part hall (Ehrhardt 2020: 438). It was therefore not particularly suitable for banqueting.
Another hypothesis was put forward by Luca Rocco. A series of 22 regularly shaped, roughly square rooms were located behind the lateral arms of the stoa on the upper level. With dimensions of approximately 4.50 × 4.50 metres each,23 and apparently lavishly decorated,24 each room could accommodate seven benches, or klinai (Leypold 2008: 153).25 However, the excavation monograph dates these rooms to much later periods (Schazmann & Herzog 1932: 14). The reasons for this dating are unclear, but apparently there were two construction phases, with only the latter showing clear indications of a more recent date (Schazmann & Herzog 1932: 17).26 However, some of the characteristics described in Herzog’s excavation diaries point to a possible Late Classical or Hellenistic date for the first phase,27 which is why these rooms were included in the model.
Assuming the functional hypothesis is correct, these 22 rooms, each equipped with seven klinai, could have accommodated 154 people dining simultaneously – a suitable capacity for the most important individuals. However, it is impossible that these rooms could have accommodated the entire festival crowd. The rooms behind the lower stoa, on the other hand, might have provided additional space for banqueting, but their dimensions were much more irregular. While this does not preclude dining having taken place there, we lack any supporting evidence.28
However, we should also bear in mind that permanent stone architecture was not the only option. There is scattered, yet convincing, evidence of tents that were temporarily erected for feasting (Goldstein 1978: 8–100). Although nothing survives from the Koan Asklepieion, the unbuilt courtyard of the lower terrace would have been the most suitable area for setting up fireplaces and cooking facilities, as well as for erecting banqueting tents for most of the festival participants (Figure 9).
Figure 9
Setting up feasting tents at the lower terrace (nightfall).
The feasting would surely have continued late into the night. Once the sacrifices had been performed and the crowd had dispersed, the staircases would have been available for vertical traffic again. Nevertheless, the cult personnel serving the upper and lower dining areas might also have used the hill slopes and lateral entrances, which were conveniently located next to the stoa ends on both levels. This would have allowed them to avoid collisions with festival participants wandering among the feasting areas.
Conclusions: Serious Gaming and the Body
Using game engines in conjunction with virtual reality glasses and a treadmill to explore and discuss virtual environments – particularly vast building complexes such as the Koan Asklepieion – allows us to approximate past built spaces from a perspective similar to that of the ancients.29 Game technology can serve as an extension of the modern user’s body and sensory apparatus, helping us to address three areas of body-related research. The first of these is the focus of the present paper, while the second and third provide insight into other fields of study within the project.
Spatial Logistics
Previous research has focused on the localisation of action sequences such as sacrifices and incubation (Interdonato 2016; Rocco 2017). Building on these hypotheses, the digitally (re)constructed built space enables us to discuss the logistics of such actions in greater detail.30 As the virtual environment presented here adheres to the laws of physical reality, this approach requires us to be much more specific. Once modelled as part of a procession, for instance, a body must be left somewhere when the next action sequence begins – whether as an observer or victim during sacrifices, or as a participant in feasting. This forces us to consider not only where sacrifices might have taken place, but also where more or less suitable gathering and observation areas might have been located – for example, the lower terrace courtyard, the intermediate terrace, or the restricted upper banqueting rooms – and which routes, passageways, or gates would have been more plausible. For example, cattle might have been led up staircases, processions might have passed through the propylon channel, and cult personnel might have travelled along fairly steep mountain slopes. Naturally, this depends on the identity and number of bodies involved in a given action sequence.
Changing Spatial Atmospheres
Earlier visualisations of the Kos Asklepieion depicted it as deserted, with the occasional figure here and there (Schazmann & Herzog 1932: Pl. 40; de Mattia 2012: 75 Figure 18). However, thanks to the ability of game technology to render moving bodies, we can now study the changing visual atmospheres that emerge from the interaction between built space and its actors. This enables us to view the human body as a significant spatial element, on par with the physical components of built spaces, such as landscapes, architecture, and sculptures. As shown in this paper, the appearance of built spaces changes dramatically when dozens or hundreds of bodies are taken into account, whether they are static or in constant motion. Furthermore, the intention is to provide a full-colour version of the model by the end of the project. This will ensure that the currently white human bodies and architecture no longer stand out against the multicoloured landscape features and cattle. Alongside the changing daylight conditions over the course of a festive day (Figures 3, 4, 5, 6, 7, 8, 9), this will provide a deeper insight into the atmosphere of built spaces during different activities.
Structuring Spatial Feeling
Previous studies of the Kos Asklepieion focused on the view of the sanctuary from a distance or from the upper terrace (Schazmann & Herzog 1932: 1–2; Gruben 2001: 448; Winter 2006: 216–217).31 However, game technology’s immersive quality allows us to discuss not only the appearance of the sanctuary or the surrounding landscape from any point of interest, but also to explore the characteristics of different areas and actions using our own spatial perception. The abrupt drop in altitude at the front of the terrace, the steepness of the hillsides and the imposing monumentality of the retaining wall near the fountain house are all important factors to consider, as is the question of what can be seen from each viewpoint. Furthermore, it is reasonable to assume that such spatial characteristics would have been at least anticipated, if not carefully planned, when the sanctuary was designed. This is evident in the road network, which is designed to direct movement vertically along the staircases and horizontally along the edges of the retaining wall between the lower and middle levels. It also directs movement outwards towards the steeply rising hillsides.
Unlike other forms of visualisation or simulation, which often convey a detached sense of scale and proportion by occasionally including small human figures, game engines allow us to investigate bodily-rooted perceptions of volume, movement and vision more experientially. Therefore, game technology establishes a more immediate connection between our bodies and past built environments, making it useful not only for the present case study, but also for other kinds of ancient built environments. The more uncertain the shapes of landscapes, buildings and bodily practices are, the greater the range of possible reconstructions becomes. However, it is precisely this uncertainty that enables game engines to demonstrate their strength; they facilitate the development of flexible and adaptable research environments that can accommodate and foster multiple hypotheses. This is practice-based research, in which the modelling process itself can lead to the transformation or development of ideas regarding ancient built spaces (Candy, Edmonds & Vear 2022: 29–30; Morgan 2022: 215).
While models generated in this way cannot be understood as direct reflections of historical reality, since they are necessarily constructed and rely on a great deal of assumptions and hypotheses to fill in the gaps in the material record,32 they nevertheless provide a powerful heuristic tool. Chrysanthi et al. 2012 have referred to this as a prosthesis. By enabling the modelling and comparison of alternative configurations of the body, architecture and landscape, game engines facilitate the systematic exploration of the implications of these configurations for movement and perception. These implications can then be critically assessed in relation to the ancient context. As the user of the virtual environment is the researcher themselves, the ideas and hypotheses resulting from such research are necessarily subjective rather than objective or statistically provable (Lanjouw 2016: 6). However, by discussing and sharing such a virtual research environment together33 (Holter et al. 2020: 226–227), using certain kinds of action as a shared reference point to compare different kinds of built spaces,34 and of course by applying the method to a greater number of ancient material records, we may in future go beyond hypotheses regarding single case studies.
Declaration of AI Use
Language was improved with the assistance of DeepL Write (DeepL 2026). The authors take full responsibility for the content.
Abbreviations
CGRN: Collection of Greek Ritual Norms, http://cgrn.ulg.ac.be/ [Last accessed 13 September 2025].
IG: Inscriptiones Graecae.
Notes
[1] There is extensive literature on this topic. See, for example: Butler & Purves 2013; ed. Day 2013; ed. Hamilakis 2013; eds. Kurapkat, Schneider & Wulf-Rheidt 2014; ed. Bradley 2015; eds. Kreuz & Haug 2016; Squire 2016; ed. Betts 2017; eds. Littlechilds & Veitch 2018; ed. Purves 2018; eds. Butler & Nooter 2019; eds. Friese, Handberg & Kristensen 2019; Aldred 2020; eds. Nadali & Pinnock 2020; ed. Ferri 2022; eds. Collar & Kristensen 2024.
[2] https://www.samothrace.emory.edu/agent-based-modeling/ [Last accessed 10 September 2025].
[3] The two authors of this contribution have written a paper that critically assesses the relationship between the altar and the upper temple within the Asklepieion of Kos, among other things. This paper is currently in press as part of the proceedings of the AIAC/CIAC 2024 conference in Paris.
[4] Project director: Asja Müller (Kiel University); technical director: Martin Kim (Mannheim Technical University of Applied Sciences); modelling: Amelie Nieberle (State Academy of Fine Arts Stuttgart) and Ecaterina Cosceeva (Augsburg University of Applied Sciences); data management: Felix Maurer (Freie Universität Berlin); archaeological process control: Danny Schultze and Tessa Rödle (both Freie Universität Berlin).
[5] This is where the project director held a position when the project started. Since April 2025, she has been employed at Kiel University, while the project itself has remained at Freie Universität Berlin.
[6] Further excavations have taken place since the 1930s. For more information, see: Laurenzi 1931: 620–621; Morricone 1950: 326–327; Rocco 1996.
[7] This suggestion was put forward by Philipp Kobusch at a workshop in Berlin in the summer of 2025. It explains both the peculiar ground plan and the fact that there is a spring house immediately behind. However, this hypothesis cannot currently be proven or disproven.
[8] A closing installation from the Post-Hellenistic period was found in the south-western corner of the lower terrace (Schazmann & Herzog 1932: Pl. 38).
[9] Competitions are not part of the modelled scenario, as there is no indication that they took place at the Asklepieion.
[10] There is good evidence that this festival was still taking place in the 2nd century BC, the period represented by the model, although evidence from this period is scarce (donation decree of Teleutias: IG XII 4, 1, 81; subscription list: IG XII 4, 2, 435).
[11] Other spatial elements that have not yet been fully included are the honorary and votive statues set up within the Asklepieion. This part is currently under construction.
[12] As for the children, no distinction in status was modelled here. Both genders wear chitons and can be identified by their smaller size compared to the adults.
[13] There is good reason to assume that the modern road to the Asklepieion largely follows the route of the ancient street, given that a line of ancient tombs runs alongside it from the city to the modern village of Platani (Bosnakis 2008: 20 Figure 1).
[15] The propylon of the Asklepieion is too badly preserved to allow us to reconstruct its former appearance. Therefore, it was modelled with four columns in antis (as suggested by Coulton 1976: 248 Figure 74; Ortaç 2001: 113) and in a style similar to the Delian propylon (Étienne 2018: 64–80).
[16] Only the embeddings of the steps are preserved. However, it was hypothesised that these steps were a similar size to those on the upper staircase (Schazmann & Herzog 1932: 22; Becker 2003: 57–58).
[17] For example, the Acropolis of Athens is only accessible to quadrupeds via the Propylaea, which proves that cattle were able to overcome entrance buildings with long and steep staircases.
[18] The hypothesis in Riethmüller 2005: 216–217 (Apollo) is thus to be dismissed.
[19] This suggestion is widely contested. Other hypotheses include the possibility that it was either Asklepios or Apollo, alongside an Attalid king (Riethmüller 2005: 216–217; Interdonato 2013: 188–189; Kobusch 2022: 215).
[20] Currently, the project focuses on visual aspects only. However, there are quite a few projects that study acoustics using digital 3D models, either through game engines or other methods (see, for instance, Schultz et al. 2017: 172–177; Fron et al. 2019; Holter et al. 2020; Kopij et al. 2023).
[21] This suggestion was made by Wolfgang Ehrhardt at a workshop in Berlin in the summer of 2025. As he pointed out, the carved steps on the slope above the Apollon Karneion terrace in Knidos would provide a good point of comparison (Ehrhardt 2014b: 32).
[22] Assuming that Becker (2003: 66) is correct in stating that the angular pedestals on either side of the central staircase landing supported statues, it would not have been possible to access the intermediate terrace from the steps.
[23] These measurements were taken from Schazmann & Herzog (1932: Pl. 7).
[24] In his excavation diary, Herzog mentions a substantial layer of red and green plaster fragments from the wall decoration that was excavated on 4–5 October 1904 (see https://digisam.ub.uni-giessen.de/ubg-ihd-nl-herzog-r/content/titleinfo/3112218 [Last accessed 11 September 2025].
[25] Rocco equated these rooms with those in the 2nd century BC Aphrodite sanctuary in Kos, which were also used for banquets (Rocco 2009: 607–608).
[26] However, during the excavation process, Herzog appears to have had a different opinion, relating the first phase to the Hellenistic period (Herzog 1905: 4).
[27] When excavating the rooms on 5–8 October 1904, Herzog differentiated between a double back wall (doppelte Rückwand) and a parallel wall (Parallelwand), see https://digisam.ub.uni-giessen.de/ubg-ihd-nl-herzog-r/content/titleinfo/3112218 [Last accessed 11 September 2025]. The parallel wall must be the back wall of the second enlargement phase of the rooms. The double back wall, on the other hand, may refer to a construction technique involving a row of small stones set against a row of larger stone blocks. This technique is evident in the late Classical to Hellenistic period, as seen in Hellenistic Building D and the rooms at the lower Stoa in the Asklepieion, as well as in the 2nd century BC sanctuary of Aphrodite in Kos City (Livadiotti 2006: 178–180; Rocco 2009: 607; Livadiotti 2013: 46–47).
[28] There are many suggestions regarding their function: cure (Schazmann & Herzog 1932: 61; Gruben 2001: 445), accommodation (Riethmüller 2005: 212), and incubation (Rocco 2017: 343).
[29] However, exploring ancient situations through the lens of one’s own body necessarily means relinquishing any claim to objectivity. It is impossible to know for certain whether our findings would have been experienced in the same way in antiquity, since movement and perception are shaped not only by biology, but also by socialisation – just like the body itself.
[30] This approach to body logistics, inferred from questions of movement and perception, complements the recent trend in sanctuary research to focus on practical activity and experience as a constituent part of sacred space (eds. Barringer, Ekroth & Scahill 2025).
[31] The current authors have a paper in print that deals with these aspects in more detail. This paper is currently in press as part of the proceedings of the AIAC/CIAC 2024 conference in Paris.
[32] This paper addresses a couple of the most important questions regarding the project’s (re)construction decisions (e.g. the altar table’s orientation within the precinct and the rooms behind the upper stoa). However, there are many more uncertainties underlying most parts of the modelled space that have not been visualised (as would be demanded by best practice in computer-based visualisation; see Denard 2012). This is because visualising them would diminish the model’s usability for the intended research questions (e.g. atmospheres). Nevertheless, in order to make the uncertainties permeating the model as transparent as possible, a fully annotated model will be published at the end of the funding period on the Refubium Platform of Freie Universität Berlin (https://refubium.fu-berlin.de/?locale-attribute=en).
[33] There is not yet an established infrastructure for researchers to meet and discuss within a particular virtual environment. However, the model resulting from the research project described here will at least be made available in a Freie Universität Berlin repository (see previous note), where it can be downloaded and developed further.
[34] In her habilitation manuscript (the second book), the first author of this paper developed a theoretical and methodological framework for analysing and comparing embodied movement and perception within different types of built space, using different types of digital 3D models. The finished manuscript, titled ‘Dynamische Räume: Handlung und Wahrnehmung in Heiligtümern des Asklepios’, is currently being prepared for publication. However, part of its framework (without drawing on digital methods, but using the related heuristic concept of the ‘avatar’) has already been explained in Müller 2024.
Acknowledgements
The article is based on a presentation that was given at the CAA conference, ‘Digital Horizons: Embracing Heritage in an Evolving World’, in Athens in May 2025.
Peer Review
A preprint version of this article has been peer-reviewed and recommended by PCI Archaeo (https://doi.org/10.24072/pci.archaeo.100636).
Author Contributions
The first author is solely responsible for creating the topic, structure and content of this article. The second author is responsible for the images and videos derived from the digital model.