Rock-Cut Monuments at Macedonian Philippi – Taking Image Analysis to the Religioscape

1 Introduction to Philippi and the Rock Monuments – History and Discovery

The geographical location and the long history of the city of Philippi make it a fascinating cultural palimpsest to analyze, given its many facets (Figure 1) (for a complete summary and discussion of the historical sources until the fourth century BC: Zannis, 2014; for the Roman colonia most recently: Friesen et al., 2022). The city was founded in inland Thrace as a colony of Thasos (itself a colony of Paros) in 360/359 BC, under the name of Krenides, which most probably alludes to the rich water sources in the area^[1] (Figure 2) (Picard, 1985, 1994; Zambon, 1999). A few years later, in 356 BC, it came under the rule of Philip II of Macedon who changed its name to Philippoi (Prestianni Giallombardo, 1999). The settlement’s location was of great strategic value, which is why the king not only secured it but also sent Macedonian citizens to inhabit it: the rocky outcrop of the Lekanis mountains, part of the southern Rhodope mountains, that is called the acropolis of Philippi nowadays, dominates the surrounding marshy plain (Figure 3), crossed by the Angites river. Mount Pangaios, rich in gold and silver, lies opposite the ancient city (Malamidou, 2019; Ungerer, 1987).^[2] And the important road that would later become, under Roman rule, the Via Egnatia, passes Philippi and transverses the plain in east–west direction. In addition, the city’s seaport (Neapolis, modern Kavala) was only about 10 km away (cf. Figure 1). The discovery of other gold mines east of the city in the Lekanis mountains (Vavelidis et al., 1997) and the partial reclamation of the surrounding land made it one of the main cities of the Macedonian kingdom, though one located in an area that still retained strong Thracian influences (up to the imperial Roman period: Brélaz, 2014a, p. 1480). Little is known about the events and transformations that Philippoi underwent following the Roman conquest of the Macedonian kingdom in 168 BC, making it difficult to understand why the city was the site, a century later, of such an important historical event as the clash between the Caesaricides Brutus and Cassius and the troops of Octavian and Marc Antony in October 42 BC (Bertrand, 2010; Butera & Sears, 2017). After the battle, the city was re-founded at Antonius’ instigation under the name Colonia Victrix Philippensium and populated with veterans of the triumvirs (Santoriello & Vitti, 1999; Tirologos, 2021); a further settlement of veteran colonists arrived in the area with the refoundation of 27 BC and the change of name to Colonia Augusta Iulia Philippensis (Brélaz, 2018a,b; Brélaz & Demaille, 2017; Brélaz & Tirologos, 2016; Rizakis, 2003, 2017; Tirologos, 2006).

Figure 1

Topographical map of the north coast of the Aegean showing Philippi and the inland plain (established by M. Hunziker using ArcGIS with a Google Satellite Image as the basemap).

Figure 2

Historical map of the area of Philippi with the marshy lake south of the ancient city (Heuzey & Daumet, 1877, map A) (München, Bayerische Staatsbibliothek; CC0; URL: https://www.digitale-sammlungen.de/de/view/bsb10221448?page=84,85).

Figure 3

View from the modern road EO 12, after crossing the Symbolon mountain ridge: the acropolis of Philippi is visible in the center (established by C. Graml with Google Earth, Streetview).

As for the material remains (Figure 4) of the city’s Roman period (Collart, 1937; Sève, 2014), the epigraphical testimonies are the only ones to have been studied exhaustively (Bartels, 2006; Brélaz 2014a,b; Pilhofer, 2009) and the best-known monument is probably the theater, which has a previous Hellenistic phase (Figure 5) (Karadedos & Koukouli-Chryssanthaki, 1993, 1999a,b, 2001a,b, 2005, 2006a,b,c, 2007); it was restored in the 1950s to function as a venue for theatrical plays. Another important episode that was to shape the city’s history is the visit of the Apostle Paul around AD 49–50, which engendered one of the oldest Christian communities – the first on European soil! – as well as the gradual spread of cult-related buildings (Friesen et al., 2022). After a period of economic prosperity in Late Antiquity, Philippi was destroyed by an earthquake in AD 619 and never rebuilt (Fournier, 2016).

Figure 4

Map of the archaeological site of Philippi (Brélaz, 2018a, Figure 4) (URL: https://books.openedition.org/efa/file/3187/tei/img-4.png/download/1500).

Figure 5

View of the ancient theatre of Philippi, seen from the south (photo: C. Graml).

The nineteenth-century interest in Greece, which had long been difficult to access due to Ottoman rule, and the city’s archaeological remains, hailed in literary sources, brought attention back to Philippi (cf. Figure 2; first archaeological mission to Macedon, published by Heuzey & Daumet, 1877): it became the destination for excavations by French archaeological missions, first in the mid-nineteenth century and then on several occasions in the first half of the twentieth century. Later, further excavations were carried out by Greek institutions and universities (cf. https://whc.unesco.org/uploads/nominations/1517.pdf, p. 155). Currently, only a small part of the circum-walled city and extra muros basilicas to the east of the city walls have been excavated. The majority of the impressive visible architectural remains date to the Roman or Late Antique period. Since 2016, the city, including the neighboring historical battlefield, has been classified as a UNESCO World Heritage Site due to its historical importance and for its Roman and early Christian architecture (cf. https://whc.unesco.org/en/list/1517/).

Some of the most remarkable ancient remains at Philippi are undoubtedly the almost 200 attested rock monuments, scattered on various rocky slopes of the mountain immediately east of the built city space (denominated as acropolis) which are the subject of the LMU project “Rock-cut reliefs in Philippi: a microregional study on the religion of ancient Macedon.” As is often the case, monuments that have always been visible tend to be less well researched and scientifically investigated (Figure 6: first illustration of 11 reliefs: Heuzey & Daumet, 1877, Pl. 4). A first study of the reliefs was presented in 1922 (Picard, 1922), while a complete edition of the reliefs, niches, and rock inscriptions was published only in 1975 (Collart & Ducrey, 1975). This study included a map of the 187 known monuments (Figure 7 ^[3]) that took stock of the visible ones as well as the already destroyed but documented ones. The locations of the reliefs, inscriptions, and niches were divided into four spatial sectors along the acropolis slopes and a number of rock-monument clusters – defined as sanctuaries – were also identified. Even though they are partially located intra muros, the urban context, meaning the use and function of the rocky acropolis slopes within the city fabric, is still unknown, having never been the subject of systematic research.^[4] The same applies to the city walls, which are an impressively well-preserved monument of Philippi (Figure 8): Phillip II is thought to have established them, but the structure, course, and development of the walls have not been systematically investigated so far.^[5] Looking at the spatial relation of the wall structures to the rock-cut monuments, the circuit wall separates various groups of these from each other, creating a divide between those intra and those extra muros that may disrupt their possible mutual relationship. Paul Collart and Pierre Ducrey have studied the monuments mainly from an iconographic and epigraphic point of view. They highlight that about half of the known rock-cut monuments depict a young huntress identified as Diana (cf. Collart & Ducrey, 1975, Index) (Figure 9), whereas the other reliefs and inscriptions instead feature other deities, such as Jupiter, Minerva/Athena, Cybele, Silvanus, Liber Pater, or Isis. They propose a dating between the end of the second century and the third century AD (Collart & Ducrey, 1975, p. 257), based on certain stylistic^[6] and epigraphic criteria, as the reliefs cannot be dated on the basis of stratigraphic contexts or other methods.

Figure 6

First illustration of 11 rock-cut reliefs from Philippi (Heuzey & Daumet, 1877, Pl. 4) (München, Bayerische Staatsbibliothek; CC0; URL: https://www.digitale-sammlungen.de/de/view/bsb10221448?page=1).

Figure 7

Map of all known rock-cut monuments along the acropolis slopes of Philippi (Collart & Ducrey, 1975, map). High-resolution image available via: https://www.persee.fr/doc/bch_0304-2456_1975_sup_2_1_342?pageId=T1_268_1.

Figure 8

Detail of the city wall, north-east of the theatre of Philippi (photo: C. Graml).

Figure 9

Rock-cut relief of a young huntress with bow (Collart & Ducrey, 1975, no. 56 – photo C. Graml).

The interpretation that is usually given (see most recently Rizakis, 2017, pp. 181–183, 189–191, 2022, pp. 47–49) is that of rather spontaneous expressions of private cult, neither organized nor controlled at the level of the city government, attended by people from all social strata both from Philippi and the surrounding areas. Given the attestation of Roman, Greco-Roman, Greek, Thracian, Egyptian, and Anatolian cults, a complex, multifaceted, and likely intersecting cultural identity can be reconstructed that was permeable to external influences and shows evidence of its individual elements influencing one another. As such, it is difficult to attribute a certain “Greek” or “Roman” or “Thracian” nature to the deities attested through iconography alone, as is done on the basis of the language used in the few attested inscriptions (all in Latin, although eastern and Thracian onomastic elements also appear: Collart & Ducrey, 1975, p. 257). For the depictions of our youthful huntress, this begs the question of whether the goddess most frequently depicted is Diana, Artemis, or Bendis, or even a hybridization of all three (for the sake of brevity, see on this question Abrahamsen, 1995, pp. 45–52; Collart, 1937, pp. 431–443; Collart & Ducrey, 1975, pp. 201–209; Deoudi, 2010; Ducrey, 1988; Heuzey, 1865, pp. 449–460; Heuzey & Daumet, 1877, pp. 80–81; Lamoreaux. 2013, pp. 140–141; Picard, 1950, pp. 25–34; Portefaix, 1988, p. 116). Moreover, some female figures depicted in the rock-cut reliefs cannot be identified as specific deities due to inconclusive iconography (Figure 10) and may not even have been intended as divine figures at all (Hošek, 2012, pp. 325–333).

Figure 10

Rock-cut relief of a standing female figure with attributes characterizing her as matrona (Collart & Ducrey, 1975, no. 115 – photo: C. Graml).

With regard to the rock-cut monuments’ spatial embedding into the urban fabric of Philippi, this complex of reliefs, niches, and inscriptions (cf. Figure 7) is still traditionally referred to as the suburban and acropolis religious center or sanctuary of acropolis (Abrahamsen, 1995; Collart & Ducrey, 1975; Rizakis, 2017, 2022; Tsochos, 2012), in which reliefs (single or in groups) of Diana and other deities are distinguished from actual small shrines excavated or built on the rock. Interestingly, some of the rock-cut monuments lie outside the city walls. Dense clusters of reliefs, with or without inscriptions and niches, but characterized by a more monumental architecture, are commonly denominated as “rock sanctuaries” and include the one of Silvanus and the “sanctuaire aux niches” at the foot of the mountain closer to the urban center of the Roman forum oriented toward the Pangaion mountains across the plain (cf. Figure 6), in the area of the ancient quarries on the higher levels of the city (Collart, 1937, pp. 454–456), and the sanctuary of the Egyptian deities halfway up the hillside on an artificial terrace (Collart, 1929, pp. 70–100). The actual sanctuaries lie within the city walls. Many reliefs do have a spatial relation to stone quarries (still of uncertain date: Collart & Ducrey, 1975, pp. 12–13) and the aqueduct, which has numerous phases (Dadaki et al., 2016; Oulkeroglou et al., 2019; Provost, 2011).

While Collart and Ducrey (1975, pp. 254–257) have proposed an interpretation of the reliefs as votive depictions dedicated to different deities and influenced by a combination of different traditions and influences from Thasos, Thrace and Asia Minor, Abrahmsen (1995, pp. 63–66) and Deoudi (2015) see them as indicators of female and male cult practice; the depiction of Diana are interpreted as commemoration of events such as birth, death and healing and expression of female cult practice, while the cult of Silvanus is seen as reference to male cult activity. In her interpretation, the presence of the cult of Diana attracted other religious manifestations to the acropolis that were dedicated to deities that had no place among the officially accepted and collectively practiced cults of the colonia, which had their place in the forum and the urban context.

Compared to this traditional interpretative framework, our project adopts a different point of view and a new style of interpretation. In the article at hand, we will mainly limit ourselves to opening up the discussion on the current state of research by pointing out some on-site observations. First, we consider it necessary to re-examine the chronology of the creation of the rock-cut monuments, which so far appear to have been squeezed into a fairly narrow chronological framework that only takes into account the characteristics of a few examples, with no regard for their more complex diachronic evolution. If the excavations in the city have so far only yielded evidence of the cults of the Roman colony, one wonders – as Mentzos (2008) has already proposed – whether it is not precisely in the mountainous area of the acropolis and on the upper terraces of the city that the most ancient cults should be sought. A first question is therefore whether it is not necessary to attribute a broader chronology to the reliefs, inscriptions, and niches while maintaining the fixed points provided by certain iconographic details and the scarce epigraphic texts. This could perhaps explain the uniqueness of the evidence given by the Philippi rock monuments, which have no (known or identified) parallels in the Mediterranean of the Roman Empire dating between the second and the third century AD, while we have some much older cases that are not totally comparable, such as (by way of example only) the three groups of rock reliefs at Akrai in Sicily or the rock sanctuary of Silvanus at Tarracina in southern Latium (Leggio, 2013; Scirpo & Cugno, 2017; Di Rosa, 2018). Such a high concentration of rock-cut monuments is in fact unique: there are no comparable contexts in Greece, Macedon, Thrace, or even Italy known so far (those proposed by Collart & Ducrey, 1975, pp. 254–255 in many cases are not comparable in terms of context, form or style or are composed of a single element). Interpreting this, as Deoudi (2010, pp. 92–94) does, as a manifestation of cults originating in Thrace at a time when Thrace was heavily romanized/under Roman control is a historical paradox, as is the attribution of “simpler” open-air religious practice to this territory, not least because some of the deities and human figures depicted in the reliefs are represented inside temple-like aedicules (Figure 11). This phenomenon, more densely attested at Philippi than anywhere else, must have a cultural–historical explanation that still eludes us.

Figure 11

Rock-cut relief of a young huntress with an accompanying dog framed by an aedicula (Collart & Ducrey, 1975, no. 12 – photo: C. Graml).

Another important issue is the spatial relationship that these reliefs have with each other, with the urban context, and with fundamental urban structures such as the theater, the quarries along the slopes of the acropolis, and the city walls or other traces of landscape use (cf. Figure 4). Carrying out a new on-site survey also made it possible to assess elements such as possible paths along and leading to the top of the acropolis, which can often be linked to processions, as well as the intervisibility of the various groupings and so-called sanctuaries; in a nutshell, the reliefs must be analyzed by placing them in a three-dimensional spatial context and relating them to the rest of the known historical topography (spatialization).^[7] With such an approach, it is evident that one cannot consider the acropolis a single open sanctuary and that it is necessary, for example, to consider separately the rock sanctuaries in the upper part of the built city, i.e., the lower acropolis slopes that are seamlessly embedded into the urban context, from the reliefs made in the ancient quarries within the circum-walled city and of course, from those made high up at various points of the rocky ridge of the acropolis, which do not seem to have been intended to be reachable and frequented at all times, but rather to be visible, probably even from quite a distance. The visibility of the reliefs then raises the question of a potential relationship with the theater: in practice, at least in its last phase, the theater screens/“hides” a large number of the reliefs to such an extent that one wonders if the additions of this last building phase postdate certain reliefs (cf. relief nos 9, 10, 107, 109, 129, 131, 134, 165, 176, 177, and 178 in Figure 7). Similarly, the reliefs are located on either side of the city walls, so if considered a rather homogenous group, they would be simultaneously urban and sub- or even extra-urban. A possible key to their interpretation might thus be to consider the reliefs on the basis of their elevation and relationship with the urban space, given that some are inside the city or on its margins, others visually dominate it, and still others, those located to the north-west near the aqueduct, are visible from afar and therefore have an entirely different relationship with the city and the surrounding area.

Finally, it seems difficult to assert that these are expressions of wholly spontaneous religiosity, unregulated by central authorities: besides the time it takes to create a relief in marble, it is necessary to consider the possibility that planning permission had to be obtained or a payment made for usage rights, since the hill must necessarily have been the property of the city itself.

To tackle our research questions and working hypotheses, photogrammetrical documentation was chosen due to its on-site practicability and to digitally and spatially document the reliefs, adding a three-dimensional perspective.

2 Methodology in Theory: Documentation of Archaeological Artifacts

Recording archaeological finds is an integral part of archaeological research and is done with much precision and detail in the recording, analysis, and preservation of finds. This important activity integrates traditional methods with modern technologies; in other words, it uses techniques like drawing and photography in combination with 3D scanning of the finds, resulting in comprehensive records (Morgan & Wright, 2018). The process requires proper planning, accurate execution, and detailed follow-up from beginning to end to ensure the accuracy and reliability of the gathered information. One of the main concerns in this respect is to anticipate possible future uses of the data. It is essential that the data are prepared in such a way that it is available for the broadest possible range of applications, including visual representation and mapping. This allows the data to be used for later projects, such as the creation of 3D-printed replicas or digital, three-dimensional, and interactive presentations. Such thorough and farsighted documentation of archaeological finds ensures that the collected information is not only useful for present research but also remains valuable in the long term for educational, research, and public interest purposes.

While the technological methods for creating 3D models are crucial, it is equally important that these models are accompanied by comprehensive metadata. Indeed, metadata ensures not just future access but also the future reusability of the 3D models, holding very important information on the equipment used to digitally capture data, significant calibration procedures performed, the software version, and the data acquisition process. These metadata should use published standards and ontologies to ensure interoperability, describing 3D models of archaeological artifacts and their contexts (Faniel et al., 2013; Homburg et al., 2021). This includes not only 3D models but also all the other digitally captured forms of documentation, be it photographs, plans, or digital drawings.

Consistency and interoperability of metadata can only be accommodated by developed standards and ontologies; these ontologies supply a structured format that outlines the different types of data and how to relate them to another dataset. The metadata makes it possible for data to be presented within an ordinary and intelligible context, making it easier to be reused and exchanged by other systems and disciplines. A good illustration of such a metadata schema is that of CARARE 2.0^[8] (D’Andrea & Fernie, 2013). This schema is designed for documenting cultural objects, including their 3D digitalizations, and comprises information regarding the archaeological artifact itself, historical and cultural information, and information regarding the technical creation of the 3D model.

This structure, with documentation, metadata, and above all ontologies, has some clear benefits. It allows different researchers and relevant stakeholders to retrieve or search for a 3D model with ease. In other words, data can be selected using preview images and detailed labeling, making them directly accessible without the need for software tailored to large 3D datasets. Not only does this facilitate internal project assessment, but it also allows 3D models that have been published or offered online to be reused. More contextual information enables the reuse of models in further or other research as well. Standardized metadata formats provide for the integration of datasets of 3D models with other datasets, and the possibility of a broader analysis and interpretation. Metadata also enables long-term preservation and archiving by ensuring that all relevant information about both the generation and usage of 3D models is available, a feature that is crucial to ensuring the reusability of data.

Documenting relevant data completely at the point of capture is important and has to be recorded in the metadata. To do so, the proper schema of metadata has to be chosen, such as those, which have associated standards, like CARARE 2.0. Ensuring that metadata and data collection practices are kept up to date will require a review and mandatory update of metadata at periodical intervals. It is, therefore, important to sensitize researchers and technicians on the need for metadata creation to ensure proper documentation.

The archaeological research and documentation campaign described above have been conducted under difficult conditions in a complex and stony-soiled landscape with embedded archaeological ruins. The difficulties imposed by the research area, in terms of the difficulty of the landscape and its limited accessibility alone, required a cautious choice of documentation procedures from the very start of fieldwork. This application, therefore, envisages an integrated approach – traditional field-survey techniques combined with modern digital documentation. Especially in the initial recording and localization of artifacts and ancient structures, the use of traditional methods, and mainly photographic documentation and manual surveying, is of crucial importance. In general, these measures allow for a detailed mapping of the site as well as the first important insights into the spatial distribution of the finds. Also, the detailed documentation of chosen objects such as the inscriptions and engravings of the objects, for example, allowed a more in-depth analysis supported by high-resolution evaluations in post-processing.

Given the prior use of cameras for documentation purposes, photogrammetry presented itself as a logical next step for creating three-dimensional models. This technique, referred to as “Structure from Motion” (SfM) (Westoby et al., 2012; Brandolini et al., 2020), enables the reconstruction of 3D models of different objects and landscapes from photographs captured from different angles and positions. SfM uses the overlaps between images to determine the spatial position and orientation of the camera in each shot and extract the geometric structure of the captured object or landscape (Bianco et al., 2018, pp. 2–4). This kind of data processing permits one to produce detailed and scale-accurate 3D models from photographic documentation without the need for other special tools. Only a camera, a scale, and an optional color chart are required. The conversion of the images into a 3D model can be done off-site. Modern photogrammetric software solutions, such as “Agisoft Metashape” or “RealityCapture,” largely automate the process and output highly detailed 3D models that offer a very accurate reproduction of physical features, allowing direct measurements of distances, areas, and volumes from the digital model (Reinhard, 2016).

The first campaign was completely based on SfM: since a camera system was already part of the standard equipment, no further systems were planned/available within the budget or transportable by two people on-site. That said, the first campaign also suggested options for future projects. In light of what has been said above, it makes sense to discuss alternatives and their combinations for future campaigns (Siebke et al., 2018), especially regarding documentation in three dimensions. In general, two areas of application are distinguished in this context: documentation of individual objects, for example, architectural fragments or intricately worked stones with engravings, on the one hand, and terrain and more extensive areas, on the other hand.

For the documentation of individual objects, both SfM and 3D scanners using structured light offer effective solutions, each with its specific advantages. 3D scanners with structured light (McPherron et al., 2009; Zhang, 2018; Diara, 2023, p. 6018) work by projecting a pattern (usually a set of lines or a grid) onto an object and capturing the deformations of this pattern on the object’s surface with one or more cameras. By analyzing these deformations, the scanner can reconstruct the 3D geometry of the object. These scanners are highly accurate in capturing the surface structures, with accuracies varying from a few micrometers to 0.2 mm according to device properties and scanning conditions, and offer high resolution, allowing for the capture of even the smallest details on the object’s surface. Certain high-precision systems can even reach accuracies below 50 μm. The manufacturers publish detailed accuracy specs so that users can choose the setup relevant to their objects. While some of them can be adapted to special requirements, stationary scanners generally require stable placement on a tripod, which is not always practical for fieldwork. They are also often quite bulky and require an external power supply as well as a laptop for data processing. On the other hand, these scanners provide particularly high resolution and three-dimensional accuracy, which makes them well-suited for capturing fine surface details. Autonomous models, such as the Artec Leo, which are operated with battery packs and hand-held, offer flexibility and faster scan speeds, making them ideal for on-site documentation. Nevertheless, they cannot achieve the accuracy of stationary systems (Redaelli et al., 2021); the Artec Leo, with a resolution of up to 0.2 mm and a measurement accuracy of up to 0.1 mm, is sufficient for many applications. In contrast, the exactness of SfM is significantly based on the quality of the captured images. Under proper conditions and with professional equipment, SfM can be accurate in the lower millimeter range. However, under poor shooting conditions, it is easy to get artifacts in the 3D data or 3D data that are missing details. A key advantage of SfM lies in its ability to generate high-quality textures. While the color information captured by scanner systems only allows for an acceptable representation of reality, the sharpness and quality of separate photographic shots are significantly superior. It would thus be very convenient to combine these methods (Inzerillo et al., 2019; Polo et al., 2022) to obtain both high-resolution geometry data from 3D scanners and high-quality color information from the SfM process. Merging both data sets will yield a high-quality final product in terms of both geometric accuracy and texture quality.

A particular focus of this project is the documentation of etchings and engravings on rock fragments, for which reflectance transformation imaging (RTI) represents a particularly interesting method (Beale & Beale, 2015; Mytum & Peterson, 2018). RTI is an advanced digital imaging technique that can help capture the surface texture of an object in great detail by taking a series of photographs under variable lighting directions. For the implementation of RTI, the target to be recorded is photographed from a single standpoint while the light source – usually a lamp – is moved to all possible pre-defined positions and angles for every shot, or alternatively, multiple light sources (e. g., in an RTI Dome, a hemispherical arrangement of light sources) can be used. The photos thus obtained are then merged into an interactive image using specialized software. The tool allows the light source to be moved virtually, bringing out surface details visible only from certain orientations and lighting angles. This method makes it possible to see fine surface details, etchings, and all sorts of engravings more clearly, especially those that might not be noticeable to the naked eye on-site. While SfM and other 3D scanning technologies are also capable of highlighting such details through subsequent filtering processes (e.g., Mara et al., 2010), the success and quality of this highlighting significantly depend on the quality of the captured geometry. The more accurate and high resolution the 3D model, the more successful these filters are at revealing small details. However, the direct capture of surface details by RTI can sometimes be more detailed and precise without requiring high-resolution 3D geometry as a basis. In general, RTI is particularly effective when the goal is to capture only inscriptions, engravings, and surface details. When it is important to document an object in its complete three-dimensional form, the use of 3D scanners or SfM is required. In such cases, however, RTI can be useful as an additional technique of documentation.

The terrain was also captured using SfM technology (Douglass et al., 2015). With the help of a camera, particularly well-visible areas could thus be accurately captured. For recording larger areas or hard-to-reach areas, the use of unmanned aerial vehicles (UAVs), or drones, represents a promising alternative (Scianna & La Guardia, 2019; Cucchiaro et al., 2020; Pepe et al., 2022). In this case, a camera is mounted on a drone that continuously takes shots of the terrain it flies across. There are even automatic systems that fly over a defined area autonomously and take pictures at regular intervals. Both methods apply the same procedure that is used in SfM, with the camera being the central element of image capture, but the aerial perspective allows for more comprehensive and uninterrupted documentation. Drones provide a flexible and cost-effective way of capturing high-resolution images of hard-to-access areas. The quality and the feasibility of shots taken will very much depend on the local weather and the legal restrictions on drone flights in the respective region.

Another efficient method for documenting terrain is laser scanning (Ullrich et al., 2002; Vozikis et al., 2004), also known as light detection and ranging (LiDAR). The LiDAR technology produces surveys of large areas with peak efficiency and high precision. A LiDAR instrument emits laser beams toward the surface of the terrain or object and determines the distance by measuring the time it takes for the laser pulses to be reflected back to the sensor from the surface (a method known as Time-of-Flight). By transmitting and receiving tens of thousands of laser pulses per second, the device can then compile a detailed three-dimensional model of the captured environment. Unlike other three-dimensional documentation techniques, LiDAR does not produce a surface geometry, but rather a point cloud consisting of measurement points. Depending on whether the LiDAR system includes a camera, these measurement points may or may not have color information. One can also convert the resulting point cloud to a surface geometry (mesh) if needed. LiDAR hardware is very versatile, as sensors can be deployed in various configurations, such as stationary setups (e.g., terrestrial laser scanners mounted on a tripod), on drones (UAV LiDAR) (Balsi et al., 2021; Masini et al., 2022), or on aircraft (Airborne LiDAR) (Opitz, 2013, pp. 14–17) for either point-based or large-scale surveys. Compared to, for instance, SfM, LiDAR can be used in any lighting conditions, meaning that the device can work in the dark of night, at sunset, or under a cloudy daytime sky. In addition, LiDAR is able to “penetrate” vegetation to capture the underlying terrain, which is particularly great for documenting overgrown or forested areas (Doneus et al., 2008; Opitz, 2013, pp. 20–23; Risbøl, 2013). Nevertheless, laser scanning requires specialized and, in most cases, costly equipment as well as professional knowledge in data processing, which limits the accessibility of the technology. Still, in most cases, LiDAR technology remains a powerful option for the precise and comprehensive documentation and analysis of landscapes and archaeological sites.

Effective documentation of archaeological findings is always done by combining technologies with their own strengths and respective limitations for use in different aspects of archaeological work. The size and accessibility of a site, the level of detail needed in the data to be acquired, and available resources are all decisive factors in choosing the appropriate method. SfM represents the most practicable solution in many cases, as this method essentially only requires the use of a camera – already a standard tool in archaeological documentation – hence eliminating the need for special equipment or other significant additional expenses. Despite the advantages of SfM, it is advisable to consider a wider range of documentation techniques for future campaigns to further increase data accuracy and quality. Integrating advanced methods like LiDAR for large-scale and precise terrain surveys or RTI for detailed capture of surface structures can contribute to a deeper understanding and more comprehensive documentation of archaeological finds. Optimal research results will be achieved by carefully weighing the respective advantages of the available technologies against the specific investigation goals and the project conditions.

3 On-Site Experiences of Documenting an Endangered Archaeological Site: Geology, Landscape, Climate, and Human Behavior

Primed with historical background knowledge on the site and its research history and equipped with the basics of how-to-do photogrammetry, the campaign of March 2023 was – like all previous research at Philippi^[9] – heavily influenced by terrain and climate.

Upon arriving at Philippi via the coastal motorway from Kavala, a landscape of extremes unfolded before us: shielded from the coast by a mountain range, the vast and almost completely flat plain stretches all the way to the snow-covered mountains of Pangaion and further north to the Rhodope mountains, close to the Bulgarian border. The acropolis hill of Philippi at the eastern border of the plan, although not as high as the snow-covered mountains, stands out markedly as a striking outpost of the Lekanis mountains, foothills of the Rhodope mountains (cf. Figures 1 and 3).

Changing perspective, from the top of the acropolis hill, the entire plain is observable and beyond the southern mountain range in the direction of modern Kavala (ancient Neapolis), even the highest mountains of the island of Thasos are clearly visible (Figure 12). This fact makes the landscape-bound historical dynamics even more comprehensible: given this intervisibility, it immediately becomes clear why the Archaic Greek inhabitants of Thasos sought to gain control of the coast (cf. the foundation of cities along the coast, such as Neapolis), and especially the fertile plain and the precious raw materials known from trade relations with the indigenous tribes (historical summary: Zannis, 2014). The acropolis rock with its undisputable advantages for controlling the surrounding areas – especially in a foreign, not necessarily friendly territory – was hence the obvious choice for a settlement in the area.

Figure 12

Viewshed analysis: from the summit of the acropolis (+15 m height for assumed fortifications), the majority of the surrounding plain is visible. In the distance, even the mountains of the island of Thasos can be seen (established by M. Hunziker using ESRI ArcGIS with Google Hybrid Image as the basemap).

Although archaeological traces of this early settlement phase at Krenides have not (yet?) been discovered, the later phases of the Hellenistic city as well as the terrain structure and geological features suggest that the slopes of the hill were never actually used as built spaces. This was certainly due to the material properties of the karstic limestone rock/marble (cf. Kokkorou-Alevras et al., 2014, p. 147; Tranos et al., 2009) (cf. Figure 13). The rather humid climate also makes the slopes a less favorable building place.

Figure 13

Geological map (a) and section (b) of the area of modern Krinides showing marble resources (Tranos et al., 2009, Figure 3a, URL: www.schweizerbart.de/journals/gr).

Nevertheless, they were included in the area protected by the city walls, since they enabled/prevented access to the hilltop. The latter’s denomination as an acropolis in the archaeological literature might give a wrong impression.^[10] Only a few beddings in the natural rocks suggest ancient buildings on the hilltop. The small number of inscriptions and etchings on the rock are of a religious nature (Collart & Ducrey, 1975, nos. 97, pp. 140–142, 166–172, 185; Höckmann, 1969–70). Thus, a shrine overlooking the city is highly likely although not conclusively verifiable. With regard to the function of acropoleis in other Greek cities,^[11] the administrative/political dimension of the space cannot be substantiated at all, which makes the city outline of Philippi somewhat comparable to Messene, another Hellenistic city foundation: there, the steep mount Ithome was included in the circuit walls, but all documented activity in the upper parts of the mountain is religious; no political activity happened there. Instead, the political heart of the city was located in the so-called Asklepieion in the lower parts of the city area. The location of the political heart of Krenides/Philippi is unclear, but at least the position of the agora of the Hellenistic foundation, when the city was renamed Philippi, is evident: it is congruent with the Roman Forum in the plain (Sève, 2022) (cf. Figure 4).

The lower parts of the acropolis slopes act as an intermediate zone between the city in the plain, where people actually lived, and the hilltop. This “interspace” was used by the inhabitants as a first step to produce building materials, only ever used locally, and eventually exploited for other raw materials. The archaeological traces of these activities, however, do not provide clear chronological markers (Kokkorou-Alevras et al., 2014, p. 147 without autopsy, mentions a late-Classical beginning). The assumed late Classical beginning at some of the quarries is certainly tied to the fact that the circuit wall built from this local stone is dated to the period of the foundation by Philipp II in 356 BCE. An earlier dating, however, cannot be excluded, since archaeological remains of the predecessor settlement Krenides are hardly known. The quarrying activity on these steep slopes and with this extremely craggy ground required technical knowledge and mastery, which could have been present among the local-indigenous people but also among the Thasian settlers. Thasos itself is famous for its marble quarries and highly skilled stone workers (Kokkorou-Alevras & Efstalphatopioulos, 2014, pp. 154–158). Moreover, Thasos could provide a link to the rock-cut reliefs, the actual object of our on-site investigation: at Thasos and at many other quarries in the Aegean (best-known is the Thasian mother-city Paros), relief decoration is attested much earlier than the Roman period (Doulfis, 2024; Gatto & van Haeperen, 2023; Kokkorou-Alevras et al., 2014). Similar to the traces of the quarrying process, the reliefs themselves hardly provide elements suitable for a stylistic or typological dating, as already emphasized above.

Despite their limited usability, in ancient and modern times, the slopes of the steep hill are still frequented for various reasons: the scenic view across the plain is attractive for picknicks and drinking as evidenced by the amounts of trash among the rocks. The rocks themselves are covered in modern graffiti and etchings (cf. Figure 14). Their excellent visibility from below also makes the rocks attractive for spraying (political) messages (cf. Figure 15). Besides these human acts of endangerment, the natural decay of the karstic rock is certainly the greatest threat to the relief. During the campaign in March 2023, we documented a rather recent rockslide (Figure 16), which had unfortunately occurred in the area “secteur 1” destroying the reliefs/niches nos. 3, 4, 21, 22, 27, 37, 38, 45, and 54 (after Collart & Ducrey, 1975) (cf. Figure 7). This troubling event underscores the high value of our first campaign and emphasizes the necessity of further documentation due to the naturally limited preservability of rock art.

Figure 14

Modern etching of a car (photo. C. Graml).

Figure 15

View of the steeply inclined terrain; in the back, a rock is covered with green plastic hiding a contemporary graffiti (photo: C. Graml).

Figure 16

Rockslide in secteur I: the monuments (Collart & Ducrey, 1975), nos. 3, 4, 21, 22, 27, 37, 38, 45, and 54 are destroyed (photo: C. Graml).

4 Three-Dimensional Spatialization: New Insights on the Dating of the Reliefs, Their Spatialization Within the Urban Setting and a Brief Outlook on the Far-Reaching Consequences of Their Re-Evaluation

In the course of the on-site examination other, hitherto unmapped areas along the acropolis slopes, but outside the city walls were also investigated. These areas were chosen based on the observations made on the reliefs in the circum-walled areas, namely the fact that the majority of the rock art attestations are situated in or in very close vicinity to ancient quarrying areas. A brief survey of the north slope of the acropolis hill, which had not been mapped by the previous survey campaigns, where modern quarries are still in use, lead to the discovery of large stone blocks, scattered along the transport road, which carry niches resembling the ones documented along the south and south-east slopes (Figures 17 and 18). Although they were not discovered in situ, these niches along the north slopes allow us to emphasize the fact that rock art at Philippi was inter alia spatially intertwined with quarrying activity. The discovery of these new niches in a disturbed context moreover underlines the fact that all reliefs, niches, and inscriptions applied to the natural rock are endangered in multiple ways, namely by the geologically determined natural erosion of the limestone rock as well as on-going human activity, be it in the form of quarrying and of expressions of human creativity (vandalism).

Figure 17

Block with a carved niche, documented north of the acropolis summit close to a modern quarry (photo: C. Graml).

Figure 18

Roughly dressed block with carvings on the side from the north slope of the hill (photo: C. Graml).

With these dangers in mind, the project “Rock-cut reliefs in Philippi: a microregional study on the religion of ancient Macedon” seeks to ensure a sustainable preservation of the site in several ways. The first objective is to produce up-to-date documentation with three-dimensional renderings that can be used to monitor the deterioration of the site by comparing them to the earlier photographical documentation. The French School at Athens provides open access to their archival materials (https://www.efa.gr/archives-en-ligne/), making readily apparent that entire areas have already been lost due to rockfalls or weathering. Secondly, the 3D models will provide future research with data that would otherwise be lost in the next decades (Figure 19), since the vastness of the area as well as the natural material and climate conditions do not allow extensive protection measures. Despite the protection status of Philippi as a UNESCO World Heritage site,^[12] the rock-cut reliefs have not been included in the fenced-in archaeological park, because the area they occupy is simply too large. Three-dimensional documentation of these archaeological testimonies is the only way to preserve them for future researchers. This method has become essential in the field of cultural heritage preservation for monitoring and detailed analysis of the degradation of such artifacts (Lercari, 2019; López-Armenta & Nespeca, 2024). The three-dimensional models also have significant applications outside academia, for instance in teaching about the ancient world: full-scale 3D prints could be exhibited at museums worldwide and miniaturized models of certain areas or the entire acropolis hill could make the uniqueness of the rock art at Philippi exhibitable worldwide but also enhance the experience on site, especially for the majority of visitors who do not have the time or the physical fitness to climb the entire acropolis hill.^[13] Moreover, virtual recreations of day- or night-time settings and events taking place at the sites of the reliefs or simulating movement toward these areas could be simulated in VR, making the Philippi rock art an immersive experience.

Figure 19

Three-dimensional model of the area (Collart & Ducrey, 1975), nos. 95, 137, 116, 62, 108, 61, 112, 44, 18, 19, 90, 105, 40, and 17 (established by C. Graml with Agisoft Metashape).

The acquired 3D data offer tremendous opportunities not only for documentation and visualization of these monuments to the public but also serve as an excellent basis for analyses that promise to improve our understanding of the rock reliefs in Philippi. Spatial analysis and visibility studies are particularly outstanding in this respect. Spatial analyses help to pinpoint the placement of the rock monuments and visualize their relations with the surrounding topography and urban structure. Using GIS, reliefs and niches can be precisely located and their distribution analyzed in relation to ancient and modern quarries. In this way, one can establish spatial data on the distribution of rock artworks and hence form further hypotheses about their function and meaning in the urban context of Philippi. Visibility studies provide another promising avenue of research. Such studies verify the visibility of reliefs both from within and from outside the ancient city. Digital terrain models simulate the reliefs as they must have appeared in ancient times and could thus have been visible. These analyses will therefore be cardinal for understanding how the ancients might have based their positioning of reliefs on some strategic choice to obtain certain visual effects or religiously symbolic messages. The visual tools represent the spatial relationships and analyses of the visibility within the data received most understandably.

Another very important step, besides the evaluation and analysis of the acquired data, is the storage of data during the campaign. All data obtained during the entire project, in particular the 3D data, are stored with the greatest care and provided with detailed metadata to ensure comprehensive documentation of the whole project. This metadata contains the data of interest about the objects themselves alongside information on the whole process of collecting data. Metadata and file formats are standardized, which provides interoperability and reuse for future scientific work. Thus, after the project, data can be included in research data portals.

Coming back to research on these religiously connotated reliefs, the 3D documentation campaign opened up new considerations and questions for the ongoing project. First, the chronological aspects of the rock art monuments became much more differentiated, since the local art production makes the use of stylistic dating a difficult, sometimes even futile task. This became especially obvious in comparison to reliefs kept at the local museum at Drama. A generally assumed Roman dating for all documented reliefs therefore seems highly unlikely. In contrast, the different clusters and their spatial setting within the urban fabric of the circum-walled city make an earlier dating for at least some of the reliefs and niches more likely. Inside the Hellenistic city walls, it becomes obvious that the extracted material was most likely used for the intensified building activity of this Hellenistic period, namely for the city walls, the agora, and the theater. With regard to the commonly accepted dating of the reliefs to the Roman period, specifically the second century CE, the several centuries between the end of activity at the quarries and the creation of the rock monuments call for an explanation.^[14] Therefore, a full-scale spatial analysis of the entire acropolis rock using 3D documentation (3D landscape models with embedded detailed models of the rock art clusters [shrines?]) and the embedding of all attested uses of this extremely difficult terrain (steep and rubble-strewn slopes) takes the research on these religious objects to a new, more comprehensive level. The temporal relation of quarrying and the creation of the reliefs demand a much closer look, to unravel the relation of the reliefs to the stone production sites and their religious meaning. By uncovering other uses of these peculiar urban spaces, the religioscape of the reliefs will in turn shed more light on the ancient lived religion at Philippi (Gasparini et al., 2020).