Stations of the Publicum Portorium Illyrici are a Strong Predictor of the Mithraic Presence in the Danubian Provinces: Geographical Analysis of the Distribution of the Roman Cult of Mithras

1 Introduction

The publicum portorium Illyrici was a Roman customs office that originated in the first century CE (Dobó, 1940; De Laet, 1949; Ørsted, 1985). Its original function was to collect portorium (the Roman alternative to customs duties) on goods imported into Italy from the Northeastern provinces of the Roman Empire. Due to the rapid development of trade and infrastructure, the conductores, private tenants of concessions for the collection of portorium, managed to extend their area of activity deeper into the Danubian provinces. The effort to extend their reach reached its peak between 160–170 CE, when the publicum portorium Illyrici merged with the publicum portorium ripae Thraciae. It continued in this form briefly until the last quarter of the second century CE, when, during the period of consolidation of the Roman Empire following the Marcomannic Wars and the outbreak of the Antonine Plague, the responsibility for collecting the portorium dues in each province passed into the hands of procuratores appointed by the emperor, mostly recruited from the ranks of the equestrian order (Van Haeperen, 2020, pp. 167–168). At the height of its activity, the customs office operated in the Ionian-Adriatic (Regio X Venetia et Histria and Dalmacia) and Danubian provinces (Noricum, Pannonia Superior, Pannonia Inferior, Moesia Superior, Moesia Inferior, Dacia, and Thracia). The daily operation of customs stations then fell to junior officials, mostly recruited from the ranks of slaves, at first privately owned by individual conductores and later selected from the pool of imperial slaves (Günther, 2015). A significant number of slaves and freedmen were also employed at the regional headquarters of the portorium in provincial administrative centers (e.g., at Poetovio in Pannonia Superior).

Members of the publicum portorium Illyrici left an indelible mark on epigraphic material, as can be seen from a large number of dedicatory inscriptions on altars, statue pedestals, or other monuments of an iconographic nature. Most of these objects were dedicated to various deities of the Graeco-Roman pantheon. Scholars of the religious history of the Roman Empire soon realized from these inscriptions that members of the publicum portorium Illyrici were actively involved in the Roman cult of Mithras (also known as Mithraism or the Mysteries of Mithras) (Cumont, 1913, pp. 73‒74). This cult, which worshiped an originally Persian (and more generally an ancient Indo-European) deity (Lahe, 2018, 2019), first appeared in the archaeological record around the turn of the first century CE (for a review of the earliest attested evidence, see Chalupa, 2016, pp. 76–89), but within a few decades it had spread virtually throughout the whole of the Roman Empire (Figure 1). This rapid and spatially impressive spread of this new religious tradition is explained by the fact that the Roman cult of Mithras was able to penetrate the highly mobile strata of Roman society (Beck, 1998, pp. 117–120). For a long time, soldiers of the Roman legions and auxiliary units were considered the main recipients and propagators of the Mithraic cult (Cumont, 1913, pp. 36–60; Daniels, 1975), but a detailed analysis of Mithraic epigraphic evidence has shown that this idea, in its extreme form, is untenable (Clauss, 1992, pp. 267–269; Chalupa et al., 2021; Gordon, 2009). In fact, a study of the social statuses of documented Mithras worshippers suggests that the cult of this deity must have been able to reach other segments of the Roman population very early on. Another non-military group that was soon identified as an important pillar of the Mithraic cult comprised members of the publicum portorium Illyrici. Their involvement in the activities of the Mithraic cult and their professional movement in the provinces where the customs office operated has thus provided scholars with another compelling explanation for the rapid and effective spread of this new cult in the Danubian provinces from the early second century CE (Beskow, 1980; Clauss, 1992, pp. 297‒299; 2000, pp. 21–23, 37–38; Merkelbach, 1984, pp. 167–174; Tóth, 1977, 1995; Will, 1970). This idea also seems to resonate with the younger generation of scholars (Egri et al., 2018, pp. 269‒270; McCarthy et al., 2017, pp. 379‒380; Mihailescu-Bîrliba, 2006, pp. 120‒124; Silnović, 2022, pp. 73‒74; Szabó, 2015, pp. 412‒413; Teichner, 2013, pp. 64‒67).

Figure 1

Overview of the spatial distribution of evidence for the Roman cult of Mithras in the Roman Empire.

However, this conclusion about the extraordinary popularity of the Roman cult of Mithras among members of the publicum portorium Illyrici was not based on a rigid analysis of all the dedications to Graeco-Roman deities made by them, but on an intuitive evaluation of the fact that some important Mithraic finds were made by persons active in the portorium and that the Roman cult of Mithras usually had a strong presence in the administrative centers of the provinces, where customs headquarters were also located. In recent years, this gap in research on the relationship between the publicum portorium Illyrici and the Roman cult of Mithras has been filled by two important studies based on a thorough analysis of a more complete set of epigraphic evidence. The first one is the article by Françoise Van Haeperen (Van Haeperen, 2020), who comes to conclusions that challenge the idea established by previous scholars that the members of the portorium publicum were a privileged group among the worshippers of Mithras and the most significant vector for its transmission and persistence in the Danubian provinces. Van Haeperen, after analyzing the inscriptions left by members of the portorium, admits that Mithras is the most represented deity in this set, but argues that the overall picture is significantly distorted by the fact that a full half of the dedications to this deity come from mithraea I, II, and III in Poetovio (Van Haeperen, 2020, p. 178). In addition to Mithras, the members of the portorium left dedications to a variety of other Roman and local deities, demonstrating that their religious preferences were diverse (Van Haeperen, 2020, pp. 178‒179). Van Haeperen further argues that Mithras was not worshipped in all places where the existence of portorium stations is confirmed, according to the extant inscriptions, and that no members of the portorium are attested among the worshippers of Mithras in various provinces where the portorium Illyrici was active, e.g., in the tenth Italian region (Venetia et Histria), Pannonia Inferior, Dacia, and Thracia (Van Haeperen, 2020, pp. 179‒180). This state of evidence compels her to conclude that

[i]l apparaît …, que Mithra ne semble pas davantage avoir fait l’objet d’une diffusion que les autres dieux honorés par les agents des douanes. La divinité perse n’était qu’une des divinités que pouvaient honorer les esclaves des douanes, sans que ce choix ne soit exclusif, sans qu’il ne soit ‘obligatoire’ …. La mobilité de ce dieu ne se révèle guère différente de celle des autres divinités auxquelles s’adressent majoritairement les agents du portorium. (Van Haeperen, 2020, p. 182)

The second major contribution to the discussion of this topic is a doctoral dissertation by Juraj Sarkisjan (2022), whose conclusions are more sympathetic to the previous views on the relationship between the Roman cult of Mithras and the portorium Illyrici, although he also offers some modifications to the overall picture. Sarkisjan meticulously collected and analyzed all dedicatory inscriptions made by attested members of the publicum portorium Illyrici (Chalupa et al., 2024) and concluded (in broad agreement with Van Haeperen) that Mithras was the most revered Roman deity in this social group. His research categorized and analyzed 126 epigraphic monuments that were demonstrably acquired by members of the publicum portorium Illyrici. Of these monuments, 36 were dedicated to Mithras, followed by Jupiter with 19 dedications. The number of dedications to Roman emperors (who may have enjoyed divine status in the eyes of the members of the portorium) and to Genius was also quite significant, each kind appearing 17 times. Of the other deities, only the goddess Isis (nine cases) and the god Hercules (four cases) appeared more frequently. The remaining dedications were to other Graeco-Roman or local deities, represented in units of cases (Sarkisjan, 2022, pp. 329–335). The high number of dedications to Mithras thus in Sarkisjan’s opinion confirms the intuitive assessment of previous research that officials of the publicum portorium Illyrici were one of the most important agents in the process of spreading the Roman cult of Mithras in the Danubian provinces of the Roman Empire. At the same time, however, Sarkisjan’s research has shown that this conclusion is only valid in a nuanced way when taking into account the regional distribution of the portorium inscriptions dedicated to Mithras. In fact, the conclusion that the collective body of officials of the publicum portorium Illyrici comprised supporters of the cult of Mithras, in general, cannot be supported on the basis of the inscriptional evidence alone. The geographical distribution of the portorium inscriptions makes it clear that the publicum portorium Illyrici was staffed by regionally active groups that probably had only limited interprovincial contact with each other and exhibited strong autonomy in their activities and religious inclinations. A significant presence of the cult of Mithras among members of the portorium can be proven by inscriptional evidence only in the territory of the Western portorium provinces (Regio X Venetia et Histria, Pannonia Superior, Noricum, and Dalmatia), with a center at Poetovio in Pannonia Superior, where 32 of the 65 identified officials of the publicum portorium Illyrici appear on attested dedications to Mithras. In the Eastern cluster of the portorium provinces (Pannonia Inferior, Moesia Superior, Moesia Inferior, Dacia, and Thracia), the intensity of the involvement of its officials in the Roman cult of Mithras was much lower: of the 44 identified members of the publicum portorium Illyrici, only 5 made dedications to Mithras (Sarkisjan, 2022, pp. 351–367).

However, epigraphic evidence is a material of a specific nature, the testimonial value of which can be negatively affected by numerous inherent distortions and biases (the preservation of inscriptions is often decided by chance circumstances; the discovery of epigraphic monuments is determined by the frequency of building activity or archaeological excavations and the nature of the sites selected as suitable or attractive for archaeological research; inscriptions are often written on a limited surface, leaving no space for additional information, such as the social status of the dedicants; people from lower social classes are generally less represented on inscriptions than people from more affluent classes, women less so than men; some inscriptions have been discovered in different places, often far from their original place of dedication and display, due to the secondary use of the objects on which they were written – for example, as building material). It is also possible that members of the publicum portorium Illyrici acquired additional epigraphic monuments, but their involvement cannot be determined due to the fragmentary state of the inscriptions, or that they preferred to dedicate anepigraphic material – for example, iconographic monuments. Aware of these practical and methodological problems, we decided to investigate the importance of members of the publicum portorium Illyrici for the spread and functioning of the Mithraic communities in the Danubian provinces by analyzing the spatial proximity between places where the customs stations of the publicum portorium Illyrici operated and places where Mithraic material of an epigraphic, iconographic and architectural nature (in the form of confirmed Mithraic sanctuaries, so-called mithraea) was discovered. This decision was based on the assumption that if the relationship between the Roman cult of Mithras and members of the publicum portorium Illyrici was indeed strong and important, the remains of the Roman cult of Mithras will be concentrated in or near places where customs stations operated. To be able to argue that any spatial proximity found between the sites of customs stations and sites with documented Mithraic finds is not coincidental but supports the thesis of close, strong, and long-lasting involvement of the members of the publicum portorium Illyrici with this religious tradition, a comparative analysis of the spatial proximity between the sites of customs stations and sites with the documented presence of other cults of selected deities corresponding to Mithras either typologically (the so-called Oriental or rather “Orientalized” deities such as Jupiter Dolichenus and Isis) or in terms of specialization (Mercurius and Silvanus) was conducted. In addition to these deities, other gods and goddesses appearing in dedications made by members of the publicum portorium Illyrici (Hercules and Jupiter) and two traditional deities of the Graeco-Roman pantheon (Apollo and Minerva) were also included in the analysis. Thus, for the purposes of this study, the following hypotheses were formulated, which we attempted to test by analyzing the spatial distance between the portorium stations and the sites of epigraphic remains of the cults of selected Oriental and Graeco-Roman deities: H1 – There is a strong spatial proximity between the sites of the stations of the publicum portorium Illyrici and the occurrence of documented evidence of the presence of the Roman cult of Mithras; H2 – The spatial proximity between the sites of the stations of the publicum portorium Illyrici and the sites of documented evidence of the Roman cult of Mithras is an order of magnitude narrower than the spatial proximity between the sites of the stations and sites of the attested presence of other cults selected for comparison, or such spatial proximity cannot be detected at all in the case of these deities.

2 Materials

To carry out the spatial proximity analysis, the following datasets were created:

1. A dataset of locations of the documented existence of customs stations of the publicum portorium Illyrici in the territory of those Roman provinces where this office was active; the identification together with the geocoordinates of the portorium stations were derived from epigraphic records attesting to (a) the location of a statio, (b) the collective activities of members of a particular portorium station, and (c) the ranks located there. This dataset was based on the selection of portorium stations belonging to the publicum portorium Illyrici and publicum portorium ripae Thraciae administrations listed by Ørsted (1985). The list of identified stations together with the associated epigraphic evidence as represented by the Epigraphic Database Heidelberg (EDH; Alföldy et al., 1997) and the LUPA database of stone monuments (UEL; Egger et al., 2002) are available at the Zenodo repository (N = 27) (Chalupa et al., 2024).

2. A dataset of sites with the documented presence of the Roman cult of Mithras in the territory of the provinces where members of the publicum portorium Illyrici were active (Mithras_AC, N = 246). The data for this dataset were collected through critical processing of the information contained in the hitherto standard corpus of Mithraic material (CIMRM = Vermaseren, 1956–1960) and from a brief survey of sites and finds of the Roman cult of Mithras in Manfred Clauss’s monograph Cultores Mithrae (Clauss, 1992). The collection of these data was followed by bibliographical research by the authors of this study, which supplemented these data with new finds made and published after 1990, and by a critical evaluation of previous data with regard to verification of the association of these finds with the Roman cult of Mithras, which led to the exclusion of many disputed finds or finds included in Vermaseren’s corpus of Mithraic material on the basis of erroneous conclusions. In addition to epigraphic evidence, this dataset included anepigraphic objects of an iconographic nature and reliably identified Mithraic sanctuaries (so-called mithraea). Only data with the attributes “confirmed” and “highly probable” (values 3 and 2 in the attribution of certainty) were used in the analysis (Chalupa et al., 2024).

3. A dataset of sites with the epigraphically attested presence of the Roman cult of Mithras and cults of selected Graeco-Roman deities was selected for comparison, also on the territory of provinces where members of the publicum portorium Illyrici were active. For the purpose of comparing data within a consistent dataset, and also for further comparison with data contained in the Mithras_AC dataset (see no. (2) above), Latin epigraphic proxies for the locations of the cults of the deities Mithras, Minerva, Jupiter, Jupiter Dolichenus, Asclepius, Apollo, Mercurius, Silvanus, Isis, and Hercules were selected from the dataset Latin Inscriptions in Space and Time (LIST v1.2; Kaše et al., 2024). LIST is a compendium of Roman epigraphy from the Epigraphic Database Heidelberg (Alföldy et al., 1997) and the Epigraphic Database Clauss-Slaby (Clauss et al., 2007). LIST contains 525,870 inscriptions with 65 attributes such as terminus post quem and ante quem, the text of the inscription, the social status of the persons mentioned, the type of inscription, etc. LIST is a representative dataset in the field of digital epigraphy for the Roman provinces relevant to this study. From the LIST dataset, and in a programming environment using Python scripts (Geopandas library), we extracted all inscriptions that mention the names of the selected deities in forms according to their declensions and known variants (deities_decline file on Zenodo; Chalupa et al., 2024). The number of inscriptions found for each deity was as follows: Mithras (N = 493), Minerva (N = 1,283), Jupiter (N = 4,031), Jupiter Dolichenus (N = 330), Asclepius (N = 388), Apollo (N = 929), Mercurius (N = 1,809), Silvanus (N = 2,333), Isis (N = 387), and Hercules (N = 1,159). Using the geocoordinates of the inscriptions from their original databases, we were able to project these data onto a map in a Geographic Information System (GIS) environment (data at Chalupa et al., 2024).

The data used in the analysis, including references to primary epigraphic sources, and the results of the measurements are deposited under an open-access license in the Zenodo repository (Chalupa et al., 2024).

3 Analysis and Results

Measuring the distance from each site with a publicum portorium Illyrici station to the nearest site with a confirmed presence of the Mithraic cult (and subsequently to sites with evidence of other deities to be compared) in GIS was chosen as an appropriate means of exploring the spatial relationships between the data. Since the stations of the publicum portorium Illyrici are grouped on the map into two separate geographic clusters, which also correspond to two different socially delineated administrative segments identified through the evaluation of epigraphic evidence in Sarkisjan’s dissertation (Sarkisjan, 2022: pp. 306–328), we assigned each station the attribute of belonging to either the Western or the Eastern cluster (Figures 2 and 3). Furthermore, in all analyses, we performed each measurement separately for the Western and Eastern geographic clusters.

Figure 2

Identified stations of the publicum portorium Illyrici in the Roman Empire.

Figure 3

Identified stations of the publicum portorium Illyrici in the Roman Empire with locations of attested Mithraic presence based on the Mithras_AC dataset.

The next step was the actual measurement of the distances from the locations of the portorium stations to the geographically closest data approximating the presence of the Roman cult of Mithras, separately for the Western and Eastern clusters. This measurement was repeated using the same procedure for the data approximating the presence of the cults of other Graeco-Roman or Oriental deities (Minerva, Jupiter, Jupiter Dolichenus, Asclepius, Apollo, Mercury, Silvanus, Isis, Hercules). The results of the measurements are shown in Figures 4 and 5. We then performed this measurement on the Roman transportation network to check whether mobility on the transportation network (measured in kilometers) would modify the results obtained in the previous step. For transportation network measurements, we used the Roman roads dataset based on the Barrington Atlas of the Greek and Roman World (Talbert, 2000), digitized by the Ancient World Mapping Center. In particular, we used a version of this Roman roads dataset with validated geometries in GIS for spatial analysis purposes provided by the ASCNET project (Glomb, 2021a; for the principles and uses of such approaches in archaeology and the study of religions, see also Kaše et al., 2024).

Figure 4

Box plot chart showing the distances from unique sites with portorium stations from the Western cluster to the nearest sites with evidence of the cults of individual deities. The Y-axis shows the length of the distance reached in kilometers. Box plot: center lines show the medians; box limits indicate the 25th and 75th percentiles; whiskers extend 1.5 times the interquartile range from the 25th and 75th percentiles; outliers are represented by dots.

Figure 5

Box plot chart showing the distances from unique sites with portorium stations from the Eastern cluster to the nearest sites with evidence of the cults of individual deities. The Y-axis shows the length of the distance reached in kilometers. Box plot: center lines show the medians; box limits indicate the 25th and 75th percentiles; whiskers extend 1.5 times the interquartile range from the 25th and 75th percentiles; outliers are represented by dots.

After the geographical distance measurements were completed, we performed descriptive statistics on the results. Since the distribution of the measured data was non-normal, as determined by the Shapiro–Wilk test for both the Western and Eastern clusters, we conducted a Kruskal–Wallis H test, which sorted the data by value into ranks and tested whether the measured distances for individual cults came from the same distribution. For both clusters (Western and Eastern), the Kruskal–Wallis H test indicated that there was a significant difference in the dependent variable between the different groups of cult proxies: for the West, χ ²(10) = 31.51, p < 0.001, with a mean rank score of 42.55 for Mithras, 72.36 for Minerva, 34.59 for Jupiter, 71.27 for Dolichenus, 86.36 for Asclepius, 70.91 for Apollo, 69.23 for Mercury, 62.64 for Silvanus, 78.32 for Isis, 54.95 for Hercules, and 27.82 for Mithras_AC (Table 1); for the East, χ ²(10) = 39.44, p < 0.001, with a mean rank score of 80.28 for Mithras, 88.75 for Minerva, 54.94 for Jupiter, 86 for Dolichenus, 119.06 for Asclepius, 83.09 for Apollo, 106.78 for Mercury, 89.5 for Silvanus, 134.72 for Isis, 83.56 for Hercules, and 46.81 for Mithras_AC (Table 2). Measurements taken on the Roman road network are entirely consistent with the distance ratios between the deities and confirm these observations (results available on Zenodo; Chalupa et al., 2024).

Although there are significant temporal uncertainties in the dates of origin of the measured data, with ranges of decades or even centuries between their termini post quem and ante quem, we checked whether the connections between the portorium stations and their nearest measured proxies for the Roman cult of Mithras (dataset Mithras and Mithras_AC) are chronologically plausible. Upon examination, all measured connections were evaluated as chronologically plausible, i.e., the dates of the origin ranges for the pairs of individual portorium stations and their closest measured evidence for the Roman cult of Mithras either overlap or the presence of a portorium station is attested slightly earlier than the closest proxy for the Roman cult of Mithras. This is true for the distances measured in both the Western and Eastern segments. The only exception is the case of the Statio Dimensis (East), the existence of which is attested approximately in the middle of the second century CE. The closest Mithraic evidence from the Mithras_AC dataset is a fragmentary altar from Novae in Moesia Inferior (CIMRM 2268/2269, about 20 km from the Statio Dimensis) dated to about 100 CE. However, the Novae altar was consecrated by a slave of the portorium conductor, and the Statio Dimensis thus was adjacent to an earlier established station in Novae already directly associated with the worship of Mithras.

Finally, we focused on the spatial relationships between the stations of the publicum portorium Illyrici and the cult of the god Mithras on the territory of the Roman province of Noricum. The reason for this step was that, for this province, we documented the presence of portorium stations along almost the entire perimeter of the borders with other provinces, providing us with a suitable opportunity for a comprehensive spatial statistical analysis of the possible influence of the activities of members of the publicum portorium Illyrici on the spread of the cult of the god Mithras at the level of a single province.

As part of this research, we asked the following additional research question: If a settlement in the province of Noricum had a portorium station nearby, did it also have the Roman cult of Mithras (and vice versa)? In other words, we were interested in whether there was a correlation between the distances from settlements along the transportation network to (a) the nearest portorium station and (b) the nearest evidence of the Roman cult of Mithras. For these purposes, we collected locations with the attributes “settlement” (featureTypes), “Roman” (timePeriods), and “Precise” (locationPrecision) from the Pleiades: A Gazetteer of Past Places database (Bagnall et al., 2006). We then identified only settlements within the territory of the province of Noricum, as well as settlements within a 30 km buffer zone along the provincial borders of Noricum, to capture the possible cross-border mobility of members of the publicum portorium Illyrici on Roman roads within the province of Noricum (N = 58). From these settlements, we then measured distances to the nearest portorium stations and to the nearest sites with attested Mithraic presence on a network of ancient Roman roads (the Mithras and Mithras_AC datasets also supplemented by neighboring provinces, distances measured in meters, N of settlements in a 10 km proximity to a road = 53). For comparison, we also measured distances to proxy data for the presence of other cults in this way. As a result, we collected data on the distances from settlements to these categories: Portorium, Mithras_AC, Mithras, Minerva, Jupiter, Jupiter Dolichenus, Asclepius, Apollo, Mercury, Silvanus, Isis, and Hercules.

After calculating the distances, we examined their correlations using Spearman’s correlation coefficient. Spearman’s rank correlation coefficient (Rs; with a range between −1.0, i.e., a perfect negative correlation, and 1.0, i.e., a perfect positive correlation) measures the strength and direction of the association between two ranked variables. It is appropriate to use Spearman’s correlation analysis to evaluate correlations when the variables do not have a normal distribution, which was the case for the distances measured in this case. For each location of a Roman settlement, the correlation analysis asks the following: how far (in meters on a Roman road) is the nearest attested portorium station, the nearest proxy for the presence of the Roman cult of Mithras, and the nearest proxy for the other selected cults of the Graeco-Roman and Oriental deities? If these distances are of similar rank, the correlation is positive.

For distances measured on the transportation network, the analysis revealed statistically significant positive correlations between distances from Roman settlements within Noricum (and in the buffer zone) to their nearest (1) portorium stations and to (2) evidence of the cult of the god Mithras and the god Hercules (Figure 6).

Figure 6

Statistically significant correlations based on Spearman’s correlation rank coefficient (Rs; p < 0.05) between the network distances of Roman settlements in the territory of Noricum (including settlements in the buffer zone) to their nearest portorium stations and their nearest proxy for the presence of individual cults. The Rs value is inside the field for each correlation.

4 Conclusions

Spatial analysis in GIS revealed patterns in the data that seem to convincingly answer the research question of whether the locations of portorium stations are a comparatively strong predictor of the presence of the Roman cult of Mithras in geographical space. According to spatial analysis of the epigraphic evidence for each of the studied cults included in the LIST dataset for the Western cluster of portorium stations, it became clear (both geographically and on the Roman road network) that the portorium sites in this cluster were significantly closer to the sites of the Roman cult of Mithras than any of the other cults compared, with the exception of Jupiter, whose cult is generally ubiquitous at sites within the boundaries of Graeco-Roman civilization. Specifically, 50% of the portorium stations from the Western cluster had their closest Mithraic inscription in the 0–5 km range, and the median was many times lower than that of the other cults (again, excluding Jupiter). The complete Mithras_AC dataset (which also included anepigraphic evidence of an iconographic and architectural nature) nuances this picture even further, reducing the median distance to 0.5 km. At the same time, the results for this dataset show that the portorium stations of the Western cluster were no farther than about 30 km from the nearest confirmed evidence of the Roman cult of Mithras, which is still the distance that the Stanford ORBIS project (Scheidel & Meeks, 2015) sees as a distance that Roman soldiers could reach in a day’s march. In comparison, for some of the other cults measured (with the exception of Jupiter), some portorium stations had the closest evidence of their presence at a distance of about 100 km (Table 1). These results seem to support the conclusion that the influence of the publicum portorium Illyrici on the spread and continuous practice of the Roman cult of Mithras may have been even more significant than it appears from the analysis of the epigraphic material alone (as done by Françoise Van Haeperen and Juraj Sarkisjan) if we take into account the fact that the influence of the members of the publicum portorium Illyrici, which is not evident in the material, may have manifested itself in other dedications as a result of secondary conversions, without these secondary connections being explicitly mentioned by the newly recruited worshippers of Mithras. Unlike other cults of the Graeco-Roman world, the Roman cult of Mithras did not celebrate public festivals that could be used to reach out to the public and attract new members. Socially, the Roman cult of Mithras operated on the basis of small and probably low-profile communities that recruited members through targeted proselytizing among friends, relatives, or people with whom existing cult members came into contact through their social and professional relationships (Aune, 1998; Beck, 2006). These newly recruited members (who themselves were not active in the publicum portorium Illyrici) may thus have been drawn to the cult precisely because of their connections to the staff of this office, without these connections being clearly traceable in the epigraphic material.

We recognize that this assertion is necessarily “speculative” since there is no direct evidence of the process explicitly mentioned in the surviving inscriptions. But this is hardly surprising, given the highly standardized nature of ancient epigraphy and the relatively limited space available on monuments for private dedications, which rarely reveal more detailed reasons for their offering. Despite this shortcoming, however, several supporting arguments can be made that at least demonstrate that the proposed scenario presents a realistic possibility of spreading the Roman cult of Mithras through social or occupational contact with members of the portorium publicum. First, the assumption that pre-existing social contacts or professional ties provide fertile ground for religious proselytizing is a crucial part of a well-established theory of conversion in the sociology of religion that can be supported by numerous empirical studies (albeit conducted on individuals or groups in modern societies) (Lofland & Stark, 1965; although various parts of Lofland and Stark’s theory were modified the importance of social ties remained relatively uncontested. The application of this theory to the Roman Empire is Stark, 1996). Second, it can also be shown on the basis of existing epigraphic and archaeological material that, at least in the case of some mithraea, there was a situation in which their membership consisted of a mixed population, i.e., of worshippers of Mithras recruited not only from the ranks of an exclusive professional group, in our case members of the portorium, but also of people who did not belong directly to the staff of this office. This situation can be documented, for example, in mithraea II and III in ancient Poetovio, where dedications by members of the portorium are successively supplemented by dedications by persons who do not mention this connection (and thus, with a high degree of probability, were recruited from a different social stratum) (Clauss, 1992, pp. 165‒169; Misic, 2015, p. 36; Van Haeperen, 2020, p. 181), or in mithraeum III in ancient Apulum, where the worshippers associated with the portorium mingled with the personnel involved in the salt trade (McCarthy et al., 2020, pp. 129‒130). A similar analogous situation, this time in the form of the coexistence of military and civilian populations, can be documented, for example, in the case of mithraeum I in Carnuntum (Clauss, 1992, pp. 158‒159; a new reading of the inscription CIMRM 1676, if correct, even testifies to the presence of a portorium slave active in this mithraeum, see Kremer, 2012, pp. 182‒183, no. 355). Third, if István Tóth’s theory is correct that the dedications to transitus, petra genetrix and fons perennis are an innovation that arose among the members of the portorium active in the Poetovian mithraea (Tóth, 1977, pp. 385‒386), then the other documented occurrences of these dedications may be either directly or indirectly related to the activity of the portorium, even if the Mithraic monuments found at these sites do not mention its direct involvement. This would make it possible, for example, to associate monuments and inscriptions referring to the transitus or transitus dei found at Carnuntum (CIMRM 1722 = HD071959), Tömördpuszta (CIMRM 1737 = HD071980), Aquincum (possibly CIMRM 1754 = HD068004 and HD068010; for the argument that they are dedications to transitus, see Tóth, 1988, p. 45), Savaria (HD071519), Sárkezi (CIMRM 1811 = HD032080), Skelani (CIMRM 1900 = HD052644), and Apulum (Egri et al., 2018, pp. 271‒272) with the activity of the portorium and thus to include those provinces which Van Haeperen considered to be devoid of Mithraic dedications demonstrably made by members of the portorium.

For the Eastern cluster of portorium stations, the spatial analysis revealed a different regional dynamic with respect to the distribution of material attesting to the presence of the Roman cult of Mithras. Here, within the epigraphic culture represented by the LIST dataset, the portorium stations no longer served as clear predictors of the presence of inscriptions dedicated to the god Mithras, as already predicted by the doctoral research of Juraj Sarkisjan. In this case, inscriptions attesting to the cults of Jupiter, Apollo, Hercules, Silvanus, or the goddess Minerva were as close to the portorium stations as those dedicated to Mithras. However, the situation changed considerably when measuring the distances from sites with portorium stations to the nearest evidence of the Roman cult of Mithras according to the Mithras_AC dataset. While for the Western cluster, this dataset only refined and intensified the distance results for the epigraphic data extracted from the LIST dataset, for the Eastern cluster these improved data significantly increased the proximity of attested Mithraic evidence to the portorium stations; thus, the Roman cult of Mithras became the strongest of all cults compared, again with the exception of the cult of Jupiter. According to the Mithras_AC dataset, the median distance from the portorium stations, in this case, was 0 km, which means that at least half of the portorium stations had evidence of the Mithraic cult directly at the station site, while, in general, no portorium stations had evidence of the Roman cult of Mithras more than 25 km away. This result can be interpreted to mean that in the provinces of the Eastern cluster closer to the Black Sea coast, anepigraphic monuments may have played a greater role, while in the Western cluster, the epigraphic tradition was more dominant and left more traces in the surviving Mithraic material.

Spearman’s correlation analysis in the case of the province of Noricum showed that the travel distances (measured in meters) between Roman cities and settlements and their nearest portorium stations on the one hand, and between Roman cities and settlements and their nearest evidence of the Roman cult of Mithras on the other hand (for both the Mithras_AC dataset and the Mithraic inscriptions extracted from the LIST dataset; Figure 6) were positively correlated with each other in a statistically significant way. In other words, in the case of Noricum and the surrounding frontier regions of the neighboring provinces, the analysis supports the conclusion that if a Roman city or settlement had a portorium station in its vicinity, then it was likely that the Roman cult of Mithras was also practiced, either directly at the site or nearby. In this context, the positive correlation for the cult of Hercules is also an interesting, if unexpected, result. However, this observation is partially consistent with the results of Sarkisjan’s doctoral dissertation, which identified the god Hercules as one of the deities appearing on inscriptions commissioned by members of the publicum portorium Illyrici.

Overall, the results of the spatial proximity analysis for both the Western and the Eastern clusters indicate comparatively close proximity of the portorium stations to evidence of the Roman cult of Mithras, although this proximity is shown with different degrees of intensity, being much stronger in the case of the Western cluster of portorium provinces than in the case of the Eastern cluster. Thus, analysis of the spatial proximity between stations of the publicum portorium Illyrici and sites with the confirmed presence of the Roman cult of Mithras yielded results that are in broad agreement with the results of the dissertation research of Juraj Sarkisjan, although he arrived at these conclusions using a different method. Thus, the results of the previous research, which established and evaluated the relationship between the publicum portorium Illyrici and the Roman cult of Mithras using the standard method of epigraphic analysis, are also supported by the quantitative spatial approach of regional dynamics.

It should be noted that the use of formal methods in historiography, by definition, always implies a reduction of some aspects of the complex phenomena studied, and since we are dealing with data that are uncertain (on a temporal scale, due to survivorship bias, etc.), the results provided will always be probabilistic observations rather than the statement of hard facts. Nevertheless, the results of our spatial proximity analysis lead to the conclusion that, in a spatial context, portorium stations, especially in the Western segment of the portorium provinces, are a strong predictor of the presence of the Roman cult of Mithras in a given locality. This seems to be broadly consistent with the views that the portorium and its members were an important factor in the transmission and dissemination of the Roman cult of Mithras. On the other hand, the results of our spatial proximity analysis somewhat weaken the power of the revisionist narrative of Françoise Van Haeperen, although they do not definitively disprove it, since it remains theoretically possible that the close proximity we detected is coincidental and caused by some other and as yet unidentified confounding factor.

5 Discussion

From the perspective of the study of ancient religions, the spread of the Roman cult of Mithras among the lower and middle-ranking officials of the publicum portorium Illyrici was a predictable situation, since it represented a viable religious option both for these officials and for their superiors, first the private tenants of the portorium and later the Roman provincial procuratores. The Roman cult of Mithras, which operated on the basis of small and exclusively male communities (Chalupa, 2005; Griffith, 2006), may have provided an important social anchor for people who, by the nature of their occupations, had to move frequently and thus repeatedly found themselves in new and unfamiliar environments where they had to live and work without the support of their families and relatives. The creation and maintenance of social ties through a common cult, whose “cells” operated in the places where the portorium stations were located, could have made it easier for these people to acclimatize to their new environment and to find new social contacts when they were forced to move. From the point of view of the tenants (conductores) and imperial procurators, then, the Roman cult of Mithras could be seen as a “religion of loyalty” (Gordon, 1972; Merkelbach, 1984) that brought together an otherwise disparate group of biologically unrelated individuals and allowed them to pursue their religious ambitions through advancement and promotions in an internal hierarchy that to some extent replicated the hierarchy of the portorium office (Clauss, 2000, pp. 39–41). For this reason, it is probably no coincidence that some conductores, such as Gaius Antonius Rufus or Quintus Sabinus Veranus, are repeatedly mentioned in inscriptions left by their subordinates in various places in several provinces (see, e.g., AE 2001 = HD067683; CIMRM 1493 = HD 013628; CIMRM 1501 = HD068777; CIMRM 1846 = HD028597; CIMRM 1491 = HD017399; CIMRM 1533 = HD071858; AE 2008, no. 1020 = HD065044). It is likely that these high-ranking members of the publicum portorium Illyrici encouraged their slaves and subordinates to participate in the Roman cult of Mithras, seeing it as a possible way to ensure the effective functioning and social resilience of the organization over which they presided and on which their social success depended. From this perspective, it is not surprising that another important social group in which the Roman cult of Mithras enjoyed considerable popularity was soldiers of the Roman army, whose lives were subject to similar pressures and who were compelled by their profession to profess similar values and pursue similar ambitions.

However, Van Haeperen and Sarkisjan have convincingly shown that the available epigraphic evidence does not support a simplistic generalization of the relationship between the Roman cult of Mithras and the publicum portorium Illyrici. Their research revealed that the Roman cult of Mithras flourished in the portorium environment primarily in the Western cluster of portorium provinces, and even then only in the milieu of groups associated with specific conductores (Sarkisjan, 2022, pp. 351–367; Van Haeperen, 2020, pp. 177–182). Our research on the spatial proximity of portorium stations to the documented Mithraic evidence supports these conclusions. At the same time, however, it suggests that the relationship between the publicum portorium Illyrici and the Roman cult of Mithras may have been closer in reality than can be demonstrated on the basis of epigraphic evidence alone, the preservation of which is subject to numerous contingencies that can greatly distort the overall impression of the interaction dynamics between these two subjects. From the point of view of our research aimed at finding the spatial proximity between the portorium stations and the sites of the Mithraic cult, it follows that in the Western cluster of the portorium provinces, the presence of a customs station predicts with a very high probability the presence of the Mithraic cult, higher than the probabilities for the cults of other Graeco-Roman deities, which in total number were much more widespread and popular. This may be the case either because an analysis of spatial proximity reveals links between the Mithraic cult and members of the portorium that are not preserved in the epigraphic record, or because of the spread of the Mithraic cult through secondary conversions, when the core group originally composed of members of the portorium in a number of places successfully transmitted the cult to different social groups no longer directly associated with the portorium, in which it took root, flourished, and later spread to other places.

The study of the spatial proximity between portorium stations and sites with the documented presence of the Roman cult of Mithras also confirmed the fundamental differences between the Western and Eastern clusters of provinces where the publicum portorium Illyrici was active. The spatial proximity analysis convincingly showed that the relationship between the portorium stations and the Roman cult of Mithras was much weaker in the Eastern Danubian provinces and that the Roman cult of Mithras was not as popular among the portorium members in this region as in its Western counterpart. This conclusion is again in line with the analysis of the epigraphic material carried out by Françoise Van Haeperen and Juraj Sarkisjan. An interesting question for further discussion is then what could have caused this difference in “receptivity” to the Roman cult of Mithras in the Eastern provinces of the portorium. On the basis of epigraphic evidence, we know that the Roman cult of Mithras penetrated this area very early, as evidenced by an inscription from Svishtov in modern Bulgaria, ancient Novae in Moesia Inferior, which was commissioned by the slave Melichrysus in honor of his superior, the portorium tenant Publius Caragonus Philopalaestrus (CIMRM 2268/2269 = HD020910). The inscription on the altar, depicting Cautes and Cautopates, two Mithraic torch-bearers, is dated to around 100 CE, making it one of the earliest reliably dated remains of the Roman cult of Mithras found to date. Nevertheless, in the following decades, the development of the cult in this area took a different course and followed a completely diverse trajectory than in the Western portorium provinces, where the cult of Mithras was able to establish a much stronger and more lasting hold among the members of the publicum portorium Illyrici. This situation is consistent with the fact that the Roman cult of Mithras always had limited success in the Eastern provinces of the Roman Empire (Figure 1). A possible explanation for this situation could be that the Roman cult of Mithras encountered greater resistance here due to the long exposure of the local population to Greek cultural, linguistic and religious influence, where other more effective and less costly means of creating social cohesion among the members of the portorium were available. Another factor that may have contributed to resistance to the Roman cult of Mithras and its relative failure in the eastern portorium provinces is the fact that portorium stations in the east were often located in large and more Hellenized cities (however, this factor cannot explain the weak connection between the portorium and the Roman cult of Mithras in Roman Dacia, suggesting that some other factors must be at play here). While the Mithraic cult probably reached the Western portorium provinces in the early second century CE through the portorium officials stationed in Aquileia in the tenth Italian region of Venetia et Histria and thus could be seen as a Roman cult and participation in it as adherence to Roman values, it may have met with indifference in the Eastern portorium provinces for the same reasons. This could be indicated, for example, by the fact that Roman emperors or Genii often appear among the recipients of dedications made by members of the portorium in the Eastern Danubian provinces, while they are completely absent from the dedications coming from the Western cluster of portorium provinces. However, this strategy of worshiping the ruling powers may have made sense in an environment where loyalty to Rome had long been expressed through worship of the goddess Dea Roma (Mellor, 1981) or, more specifically, the Roman emperors. Such worship was more intense in the Eastern provinces than in the Roman west, as it may have continued a tradition of monarchic worship that had already taken root in this region at the time of the founding of the Hellenistic dynasties that emerged after the death of Alexander the Great (Price, 1984).

Methodologically, the research presented here is another demonstration of the significant potential of quantitative spatial analysis in the study of regional religious dynamics in different areas of the Roman Empire (see, e.g., Glomb, 2021b; Mazzilli, 2022).

Abbreviations

AE

L’Année épigraphique

CIMRM

Vermaseren, M. J. (1956–1960). Corpus Inscriptionum et Monumentorum Religionis Mithriacae I–II. Den Haage: Martinus Nijhoff

HD

Epigraphic Database Heidelberg

UEL

Ubi erat lupa