The transformation of Mithraea in the Late Roman period

Abstract Discussions of mithraea tend to emphasize their uniformity. While it is true that many earlier mithraea do adhere to an established plan, there are a notable number of mithraea dating from the late 3rd c. onward that do not. This article discusses these various atypical mithraea, how such alterations to the standard mithraeum plan might have impacted on Mithraic rituals, and how this might have affected the experiences of the participants. It also explores why such changes occurred, observing that while in some instances this may have been to accommodate alterations to ritual practices, in others it was likely due to more mundane issues, such as limitations on space and environmental factors. The article concludes by reflecting on the implications this has for the identification of mithraea in the archaeological record.

It has been conventional to emphasize the shared aspects of Mithraic communities across the Roman Empire.This tendency may be traced back to the works of the founding father of Mithraic studies, Franz Cumont, who hypothesized that the cult, along with various other "mystery cults," was a direct import from Persia. 1 Cumont's thesis has long since been proven incorrect, but the characterization of the Roman cult as a relatively monolithic entity has endured.This is partly due to the persistence of the misleading term "Mithraism," which suggests the cult's followers had a set of rules to which they were supposed to adhere, despite there being no extant Mithraic doctrine (and there is still no evidence to suggest there ever was one).Recent studies, however, have increasingly highlighted the considerable variation that existed in Mithraic iconography, rituals, and membership. 2 One theme that has attracted little attention is the variability in the layout of Mithraic temples ("mithraea").Instead, the uniformity of certain aspects of mithraea tends to be emphasized, such as the arrangement of the benches either side of the central aisle leading to the cult image, albeit with acknowledgement of their varied decoration and settings. 3hile it is certainly the case that mithraea remained generally consistent in their layout across the 2nd c., in a recent study I observed that there were mithraea in 4th-c.Rome, Syria and Dalmatia that did not contain the typical plan.However, the number of atypical mithraea is much higher than I than supposed.The data presented here shows that over a third of the ca.24 mithraea known to have been created from the late 3rd c. onward did not conform to the typical plan. 4In some instances, I have challenged my own previous assumptions that certain mithraea contained two benches when the evidence does not support this.One example, described below, is the mithraeum at Lentia, which probably did not contain two benches given how difficult it would have been to accommodate them within such a small space.In other cases, such as at London, I accepted that the mithraeum was converted into a temple for a different deity in the early 4th c., but on reconsideration the evidence now strongly suggests a continuation of Mithraic practices within an altered architectural arrangement. 5New discoveries may also be added to the list, notably the Mithraeum of the Colored Marbles, the latest mithraeum identified at Ostia, which contained just one bench.This article documents this variation and also explores the impact such a design would have had on the experiences of the Mithraic worshippers. 6shall begin by providing a brief introduction outlining what is expected of a "typical" mithraeum and the rituals this may have facilitated.I shall then discuss the various mithraea that do not adhere to this "typical" design and how this would have impacted on the experiences of the Mithraic worshippers.Finally, I shall discuss the question of why such anomalous mithraea began to appear in the later Roman Empire.
A "typical" mithraeum Many so-called Greco-Roman mystery cults involved some form of public ritual, such as festivals and processions.The cult of Magna Mater had the Megalesia, the cult of Isis had the Navigium Isidis, and the Eleusinian Mysteries had both the Epidauria and the procession to the Eleusinion. 7In contrast, the cult of Mithras was almost unique in the extent of its inward focus. 8There is no evidence that the cult engaged in any public events, with all Mithraic rituals apparently conducted inside mithraea. 9The other aforementioned so-called mystery cults also had large, ostentatious temples that often contained at least some areas open to the public. 10Mithraea, by contrast, were usually relatively small, with space to hold only 50 initiates or fewer, and entry was presumably limited to members of the cult.There is also no evidence for mithraea with elaborate exterior decoration, suggesting this was considered of little consequence. 11thraea first appear in the archaeological record around the late 1st c.CE. 12 From this time until the late 3rd c., the inner chambers of mithraea largely adhere to a set ground plan, regardless of their size or location: a central aisle flanked by parallel benches that terminated in front of a niche or plinth where the cult image of Mithras stabbing the bull (the "tauroctony") was placed (Figs.1-2). 13The ceiling would usually take the form of a vault and was sometimes decorated to be rock-like in appearance.These rooms were usually windowless and artificially lit using candles and oil lamps.The intention of this design was to create a cave-like atmosphere; this was made explicit in Italy, where mithraea were typically referred to in inscriptions as spelaea (caves). 14In some instances, mithraea were installed in actual caves or attached to rock faces, into which the cult niche or the tauroctony itself would be carved, although the majority were constructed as standalone buildings or installed within pre-existing structures such as private houses, bath-complexes, or public buildings.
According to Roger Beck, the inner chamber was not intended to be seen as a cave alone; rather, the "cave" served as an allegory of the cosmos.Beck's model draws on the account of the 3rd-c.philosopher Porphyry: Similarly, the Persians call the place a cave where they introduce an initiate to the mysteries, revealing to him the path by which souls descend and go back again.For Eubulus tells us that Zoroaster was the first to dedicate a natural cave in honour of Mithras, the creator and father of all… … This cave bore for him the image The traditional interpretation that the cult relief depicts a sacrifice has been challenged in recent years, see Palmer 2009 and Faraone 2013.14 Clauss 2012, 48; Bricault and Roy 2021, 205-8.That in the provinces mithraea are typically referred to in inscriptions as a templum, while in rare instances they are also called aedes and sacrarium, is perhaps an indication of a divergence in how they were conceived of at an early stage.
of the cosmos which Mithras had created, and the things which the cave contained, by their proportionate arrangement, provided him with symbols of the elements and climates of the cosmos. 15w this star-map was projected onto the inner chamber of the mithraeum and the role this played in Mithraic rituals has been discussed at length in various publications by Beck and Richard Gordon. 16 There is not the space here to delve into the complexities of these hypotheses, but a basic outline is required.The cave has sometimes been used as an allegory of the cosmos due to its (idealized) shape: if one stands in a cave and imagines that the domed roof above them is reflected below their feet then they would be standing at the center of a sphere, which is the shape of the cosmos.Having created a cave-like space inside the mithraeum, a map of the cosmos was then projected onto this space via the arrangement of fixtures (the benches, aisle, niches, etc.) and symbols (the cult image, depictions of the torchbearers, signs of the Zodiac, representations of the planets, etc.).Thus, to move across the mithraeum was to move across the cosmos. 17 According to Beck, this map of the cosmos played an important role in Mithraic initiations.These initiations evidently varied in their details across space and time but were usually intensive processes, with the initiate, stripped, blindfolded and with hands bound, brought before the head of the Mithraic community: the Father.As part of the ritual the initiate would "die" and be "reborn" as a member of the congregation, and in Beck's model, the topography of the inner chamber of a mithraeum created a pathway for their 16 Gordon 1976; Beck 1988; Beck 2000; Beck 2006.17 Beck 2006, 102-15; Panagiotidou 2017, 96-105.This explains why the exterior of mithraea were not decorated, as neither a cave nor the cosmos has an exterior.souls to ascend and descend through the cosmos. 18In this model, initiates underwent several initiations as they advanced through the various Mithraic grades, with the ritual somewhat different each time.As each new grade was attained, so too was a greater understanding of the Mithraic Mysteries, with the initiate progressing through seven planetary spheres that were linked to each grade (from lowest to highest: Raven/ Mercury, Nymph/Venus, Soldier/Mars, Lion/Jupiter, Persian/Luna, Sun-Runner/Sol, and Father/Saturn).However, while the basic premise of a windowless room consisting of a bench-aisle-bench plan that served as a general representation of the cave/cosmos could be easily understood and repeated (particularly given the relative ease with which it could be constructed), caution must be urged in assuming that all mithraea were designed with the intention of containing a detailed star-map.Moreover, even in cases where the original design of the mithraeum did contain a star-map, this may not have been consistently understood by subsequent generations of Mithraic worshippers, particularly without texts to guide them.We must also remember that Mithraic communities were not all (and were perhaps even rarely) populated by men with a detailed understanding of and interest in astrology.Certainly, while mithraea constructed across the 2nd to 3rd c. adhere to a set plan, there is often variation in how they were furnished and decorated, and many have not produced any evidence of astrological symbolism. 19As Gordon has argued, while there was evidently a Mithraic rubric that the community's founder was expected to follow (e.g., the tauroctony, the "typical" mithraeum design, the Father as the leader of the community, etc.), there was ample scope to execute this in a way that reflected the varied beliefs and aims of each Mithraic community who made a bricolage of ideas, rituals and imagery circulating at the time. 20ile there may have been considerable variety in how the interior of mithraea were understood on a symbolic level, there was evidently consistent practical use of them as dining spaces. 21Feasting was clearly a common Mithraic ritual, as large faunal assemblages have been found in and around many mithraea, often including concentrations of poultry and adolescent pig. 22Drinking was also a prominent aspect of such events, with ceramic drinking vessels often bearing inscriptions in honor of Mithras. 23Feasts were perhaps held to consecrate mithraea, and probably took place following initiations and to mark certain events in relation to the movement of celestial bodies (such as the solstice). 24The feast of Sol and Mithras depicted in several reliefs also appears to have been re-enacted by the initiates, with the Sun-Runner and the Father acting as surrogates for the deities respectively. 25t is unlikely that participants in these feasts were seated at random.The Father and Sun-Runner perhaps sat closest to the cult relief, maybe in the center of the aisle, echoing the placement of their icons in the Felicissimus mithraeum, while the other grades would then recline on the benches in an order befitting their grade seniority and/or perhaps the required position of their related planet. 26As Anja Klöckner has observed, the tauroctony (and by extension the Father and perhaps Sun-Runner) occupied the central position during these feasts not only in terms of space, but also in terms of value and prestige, for it was correct Roman table etiquette that the guest of honor should take the central place in the triclinium. 27 summarize, the design of the mithraeum served three roles: A) a dining room; B) a replica of the cave in which the tauroctony takes place; and C) perhaps a map of the cosmos, to which Mithras's sacrifice gave birth and through which the souls of the initiates could ascend and descend.Of these, the reasons for A and B would be self-explanatory to any Mithraic initiate.In contrast, C is more complex and arguably would be the hardest to sustain a consistent and nuanced understanding of across multiple generations.The general idea that the cave symbolized the cosmos could perhaps be maintained, but that it was intended as a detailed "star-map" seems a more difficult proposition.Additionally, A and C would be most affected by an alteration to the bench-aisle-bench layout, such as the removal of one or both benches, for this would alter the seating arrangements of the feast (including the position of Mithras) and so have an impact on the experience of the participants.It would also now be impossible to conceive of the space as a star-map.Such changes might also have an impact on initiation rituals, with mithraea no longer able to facilitate an initiate's ascent and descent through the heavens.Of course, as noted, given the complexities of the star-map model, it is possible that in many Mithraic communities this "movement of the soul" was not part of the initiation process.What we can say with certainty, however, is that initiations which did involve this aspect could not have occurred in mithraea with atypical plans.

Mithraea with atypical plans
In multiple locations across the Roman Empire, mithraea appear from the late 3rd to 4th c. that do not adhere to the typical bench-aisle-bench plan (Fig. 3).At Hawarte in Syria, a mithraeum was installed in a cave that was accessed from the south by several steps. 28The inner chamber (6.45-7.20 × 4.80 m) lay in the northeast corner of the cave and was partitioned off from the rest of the space by man-made stone walls.The mithraeum appears to date to the end of the 3rd c., and in the mid-4th c. it underwent alterations when a podium was added in front of a niche and the first of five layers of painting covered the walls of both the ante-and inner chambers. 29In the inner chamber there were 26 Panagiotidou 2017, 99.Based on the images from the Felicissimus mithraeum, it appears that the Ravens would act as the primary servants while the Nymphs would tend to the lighting and the Soldiers food preparation, and the Lions would oversee the fire and burning of incense; see Chalupa and Glomb 2013; Gordon 2013c.Images of Mithraic initiates reclining on benches have been found at Stockstadt (Vermaseren 1960, no.1175) and at the Barberini Mithraeum in Rome (Vermaseren 1956, no.390).Gawlikowski 2000; Gawlikowski 2001; Gawlikowski 2007; Gawlikowski 2020.29   The floor of the mithraeum consisted of two earthen deposits, with the earlier stratum containing a coin of Diocletian and the later stratum containing coins dating to the 4th c.Coins of Constans and Constantius II were found under the podium.The southern bench of the mithraeum partially covers a deposit containing the remains of a ritual feast that dates to the 1st The transformation of Mithraea in the Late Roman period benches running along the east and south walls (ca.6.86 m N-S and 4.04 m E-W).No sculptures or reliefs were found in the mithraeum, but a niche in the north wall of the inner chamber presumably once held such an item.
Many of the extant paintings at Hawarte, which are part of the fifth and final composition that likely dates to the early 5th c., exhibit a mix of typical Mithraic iconography along with some images which are highly unusual.The north wall of the antechamber was decorated with scenes of dark-skinned individuals being eaten by lions, while flanking the entrance to the inner chamber are depictions of two figures standing in front of white horses.The better preserved of these two horsemen holds a short, two-headed, dark-skinned figure by a chain. 30In the inner chamber, beginning to the right of the niche a continuous frieze can be followed around the room that depicts: 1) Zeus battling snake-headed monsters; 2) Mithras emerging from the rock; 3) Mithras in a cypress tree with Sol looking on; 4) the tauroctony above symbols of the seven Mithraic grades; 5) Sol kneeling before Mithras; 6) a mounted hunter (Mithras?)chasing deer, a boar, and a panther; and 7) a city wall adorned with demonic heads with rays of light descending on them.The interpretation of, and the inspiration behind, the paintings of the hunting scenes, the riders, and the heads lining the top of a fortification has been cause for debate, with suggestions that these images drew on Zoroastrian and Manichean concepts. 31It is not within the scope of this paper to discuss these interpretations, but given these unusual images, this was evidently a Mithraic community that had been influenced by other religious movements operating in this region.
It is also noteworthy that the painting of the tauroctony at Hawarte does not conform to the typical placement in relation to the benches, as it is above the longer bench on the eastern wall.However, the image does face the entrance to the chamber, as is usually the case in mithraea.It has been suggested that the cult niche would have held another image of the tauroctony, but this is problematic as it would create two distinct focal points in the mithraeum. 32In other cases when more than one tauroctony was present, they were generally placed together, seemingly to avoid such a problem, as was the case at Dura-Europos, Poetovio, the Castra Peregrinorum in Rome, and Vulci. 33Difficulties remain even if the image placed in the niche was not a depiction of the tauroctony.While the mithraeum contained an unusual bench arrangement and iconography, this Mithraic community appears c.CE, and it has been suggested this represents the earliest phase of Mithraic activity in the cave.While this is not impossible, there is little to indicate any substantial activity in the caves during the 2nd or 3rd c.; thus, even if this is correct, the mithraeum as it was discovered is a Late Roman installation.

30
The rider figures flanking the entrance have been identified as Mithras (Gawlikowski 2007, 353), although an alternative interpretation is that they are representations of the torchbearers/ Dioscuri, who are often found flanking the end of the aisle near the entrance to the inner chamber in mithraea, see Dirven 2016a.31 Gawlikowski 2007, 352-61; Gordon 2001, 106-16; Dirven 2016b; Gordon 2017, 321-23.32 Gawlikowski 2007, 348.33   Dura-Europos: Adrych et al. 2017, 39-60; Poetovio (mithraeum III) : Vermaseren 1960, nos.1579 and 1589; Castra Peregrinorum: Lissi-Caronna 1986; Bjørnbye 2007, 29-31; Vulci:  Sgubini Moretti 1979, 269-76.There are some examples of additional tauroctonies placed in other locations around mithraea, such as at Koenigshoffen, where a small tauroctony was placed on the south wall near the entrance to the inner chamber, while the main relief stood against the back wall; see Vermaseren 1956, nos.1358-59.The transformation of Mithraea in the Late Roman period to have practiced ritual feasting in a manner similar to their counterparts elsewhere, as the faunal assemblage contained a relatively high concentration of chicken bones, although pig bones were only found in small quantities. 34It is perplexing that those participants in these feasts who sat on the bench along the east wall would face away from Mithras while they ate.The only explanation is that the Father, and possibly Sun-Runner, occupied this bench, given that they usually appear to have sat in front of the tauroctony in other Mithraic communities.As this is the longer bench, however, it presumably accommodated several people.Based on the available evidence it is difficult to explain the arrangement of this mithraeum.
The position of the tauroctony in relation to the benches would also be problematic in mithraea that contained just a single bench.In Dalmatia, the mithraea at Konjic and Jajce have each produced evidence for only a single bench; in both cases this was positioned to the viewer's left when facing the tauroctony. 35Both were constructed in the late 3rd to early 4th c., although Konjic (9.00 × 5.00 m) was a free-standing structure, while Jajce (7.00 × 2.80 m) was built against a rock face onto which the tauroctony was carved.At Konjic, the excavators suggested that another bench might have existed, but no evidence of it has been found.The approximate findspot of an altar in the southeast corner also provides an obstacle to this interpretation, as this would have encroached on where the bench stood.Again, both mithraea produced evidence for ritual feasting, although to what extent these faunal assemblages resembled those found in other mithraea is unclear. 36In any case, the position of the tauroctony to the left of the participants would not put Mithras in pride of place during the event.
The 4th-c.Mithraeum of the Colored Marbles at Ostia, currently the latest of the ca.20 mithraea found in the town, also contained a single bench. 37In the mid-4th c., the mithraeum was installed in a room attached to a former tavern, and it remained in use until the 5th c.It is possible that the whole building had been converted for cultic use at this time, with objects identified as an Isaic "crown" and the handle of a possible sistrum found inside.Graffiti interpreted as depicting the Navigium Isidis festival and another that states "To the unconquered god Mithras and to the great god Kronos" were present in the central room.The mithraeum was at a lower level than the rest of the building and was accessed by stairs from the southwest.A well, which contained fragments of ceramics usually used as tableware and storage vessels, was also present in the inner chamber. 38The floor was decorated with opus reticulatum (hence the name of the mithraeum), while the walls were painted in a basic manner designed to emulate marble.It measured 7.2 × 3 m and could accommodate around 12 people.At the far end of the chamber was a niche, presumably intended to hold an image of the tauroctony.In contrast to the mithraea at Konjic and Jajce in Dalmatia, the bench was to the viewer's right when facing the tauroctony, although again the same problem arises: the image of Mithras would not have taken center-stage; rather, it would be relegated to the periphery during cult meals.34 Gawlikowski 2007, 350.35   Konjic: Patsch 1899; Vermaseren 1960, nos.1895-99.Jajce: Sergejevskij 1937; Vermaseren 1960,  nos.1901-5.36   Gaidon-Bunuel and Caillat 2008, Table 7.
37 David 2016; David 2017; David 2018; David 2020.38   None of the articles published thus far have noted whether any faunal remains were found in the mithraeum.
There are also mithraea that have not produced any evidence for benches. 39In Rome, no benches were found in the small (3.7 × 2.4 m) early-4th-c.mithraeum at Via Giovanni Lanza 128. 40It was installed in what appears to have been a storage room in an affluent 4th-c.house.It contained a small tauroctony sat on a shelf with an upside-down column base underneath it that presumably served as an altar.The entrance to the mithraeum lay behind a small lararium, also from the 4th c., which contained a large statue of Isis-Fortuna, along with smaller images of the Lares, Serapis, Hercules, Horus-Harpocrates, Aphrodite, Dionysus, Apollo, Hecate, and Cybele.A series of steps, which were divided by a landing on which statues of the torchbearers stood, led down to the mithraeum.In some respects, this is similar to the typical placement of images of the torchbearers in other mithraea, at the start of the nave close to the entrance to the inner chamber.It is unusual for the torchbearers to appear outside the inner chamber, although this does echo the placement of the paintings flanking the entrance to the inner chamber at Hawarte, which possibly depict the torchbearers as riders.
Elsewhere in Rome there is the enigmatic structure located on the Aventine between the church of S. Saba and the Via Salvator.It consisted of a long corridor (ca.20.00 × 2.95 m), to the western side of which were three niches, with the central niche smaller than the other two.The floors within the two larger niches were decorated with geometric patterns.A door opposite the most southernly niche opened into a structure of which only a few remnants survive but which evidently included a large rectangular basin (6.40 × 4.85 × 1.95 m deep).A column was situated on each corner of the basin, with a series of steps leading down into the basin on its south side.Architecturally there is nothing to indicate this space served a Mithraic function, but its use in the 4th c. as a location for Mithraic worship has been tentatively assumed on the basis of a small tauroctony found there. 41The exact find spot of the relief was not recorded, however, and it appears very worn; thus, it may have only been deposited after it had ceased to be used as an item of worship. 42r were any benches found in the mithraeum at Lentia (Linz), which was constructed after 275 CE. 43 The mithraeum was built from spolia erected on the remains of an earlier building, and its neighboring structures have been identified as temples to the Capitoline Triad and the Imperial Cult.It consisted of several rooms, with an inner chamber that was unusually small (5.10 × 2.70 m).The mithraeum contained two small circular tauroctony reliefs and an altar dedicated to Mithras.There was also a terracotta plate bearing a dedication to IOM and fragments of a votive offering that bears similarities to those found in temples to Jupiter Dolichenus.No evidence for any benches has survived.It has been suggested that they were made of wood and destroyed in the fire that seemingly 39 There is also no record of benches being present in the mithraeum located in a cave ("Tomina Jama") at Epidaurum near Močići.No excavation of the site has ever been undertaken; this Mithraic community was evidently anomalous, however, given that they carved an image tauroctony above the entrance to the cave, making it visible to people on the outside.It has been tentatively dated to the late 3rd to 4th c.: see Evans 1883, 20-21; Vermaseren 1956, no.1882;  Bijađija 2012, 81-82; Adrych et al. 2017, 65 n.10; Silnović 2022, 159.40 Vermaseren 1956, nos.356-60; Gallo 1979; Bjørnebye 2007, 50-51.41 Gatti 1925, 383-87; Vermaseren 1956, nos.464-65; Bricault and Roy 2021, 217.42   Many thanks to Lia Ceravolo of the Sovrintendenza Capitolina ai Beni Culturali for supplying me with a photo of the relief.43 Vermaseren 1956, nos.1414-21; Karnitsch 1956.
The transformation of Mithraea in the Late Roman period marked the end of the mithraeum; however, the inner chamber was only 2.70 m wide, which if divided into a tripart bench-aisle-bench layout would leave only 0.9m for each section. 44Rarely are the width of benches in mithraea less than 1.5m, presumably to provide enough space for cult adherents to recline on; thus, the available space inside the inner chamber at Lentia makes the presence of two parallel benches unlikely. 45Indeed, of the mithraea discussed thus far, the interior chamber of the Lentia mithraeum is only larger than that of Via Giovanni Lanza.Moreover, the plan of the mithraeum depicts a burning deposit localized around where the cult image would have stood but not along the side walls.The mithraeum at Lentia is also unusual in that little evidence for meat consumption was found, but there was a notable quantity of fruit, nuts, and wheat in an adjoining room. 46sewhere, at Mackwiller in Germany there is evidence for a mithraeum that, at least in its final phases of use, had an appearance similar to that of a Romano-Celtic temple (Fig. 4).The foundations (ca.10.00 × 8.19 m) of the building were constructed from ashlar blocks, inside which was a basin that was fed from a nearby spring. 47On the eastern side of the ashlar structure, the remains of a smaller structure with drystone foundations were also found.Various fragments of Mithraic sculptures were uncovered within the ashlar structure, the earliest of which were dated stylistically to the mid-2nd c.It was presumed by Jean-Jacques Hatt that the drystone building had been a shrine to the god of the spring, while the ashlar structure was part of a mithraeum.At the end of the 3rd c., the drystone structure was burnt down, while the ashlar structure, which also experienced fire damage, was refurbished and a drystone wall was erected around the central basin.The ashlar structure was later destroyed, in or just after 352 CE, at which time a ceramic vessel containing 404 coins was smashed on the floor, while other coins were found around an altar and the central water-basin.Various fragments from Mithraic sculptures were also deposited on the floor of the ashlar structure, many of them found against the wall opposite the entrance.A new building was then erected over the ashlar structure, which Hatt interpreted as a shrine to a spring deity.The terminus post quem for the destruction of the final building is the late 4th c. 48iginally it was believed that the ashlar structure was the anteroom of a mithraeum as its form does not follow the established plan for an inner chamber, at least following the refurbishment conducted at the end of the 3rd c.However, there was no evidence for an adjoining room that contained two benches and a nave.Hatt hypothesized that the inner chamber was likely to have been dug into the rock of a nearby cliff face, although he acknowledged that if no such space was to be found (which to date it has not), then the square ashlar structure must be considered the inner chamber of a mithraeum. 49It also would have been unusual for so many fragments of Mithraic sculpture to have been deposited within an anteroom, as sculpture fragments recovered from mithraea tend to be found in the inner chamber.Moreover, the fragmentation of the sculptures and the 44 Karnitsch 1956, 194.45 Nielsen 2014, 161.46 Vermaseren 1960, no.1421.47   Mithraea were often erected over or adjacent to springs, such as at Carrawburgh, Septeuil, Poetovio, and Sarrebourg; thus the placement of the ashlar structure at Mackwiller would be typical for a mithraeum.deposition of their remains need not indicate the end of the use of a structure as a mithraeum.It is now evident that even in a fragmented state, Mithraic sculptures and ceramic vessels were still valued and retained by initiates.This has been demonstrated at Bornheim-Sechtem, where pieces of the same vessel were deposited in a shaft in the mithraeum at different times, while fire-damaged Mithraic sculpture fragments, seemingly retrieved from another mithraeum (there is no evidence for fire-damage to the Bornheim-Sechtem mithraeum), were stored in a niche. 50At Mainz, a ceramic vessel was ritually fragmented and deposited in the mithraeum, and similar evidence has been found at Tienen, Frankfurt, and Aquincum (Budapest). 51e deposition of sculpture fragments also echoes how the remnants of Mithraic feasts might have been buried in and around mithraea.As Zena Kamash has suggested, "it may well be that breaking up the statues and burying them was a powerful act of forgetting to remember… In this way, the burial of the sculpture might be seen as a corollary of the deposits of chicken and pig bones buried in the floors." 52Evidence for the destruction of cult images by a cult's own adherents is not uncommon, particularly as part of a ritual of closure, as Philip Kiernan has observed.Structured depositions dating to the mid-2nd to mid-3rd c. at Le Bernard, for example, included fragments from various sculptures, while at Calès-Mézin two statues of Jupiter were buried outside a temple in the mid-2nd c., and at Kottenheim a hand from a cult statue was buried under the floor of the temple. 53onsequently, the evidence suggests that for at least the first half of the 4th c. the ashlar structure at Mackwiller hosted the worship of Mithras but appears closer in plan to a  Ulbert et al. 2004; Wulfmeier 2004.51 Ulbert et al. 2004, 364-68.For the possible importance of ritual fragmentation in the cult more generally, see Croxford 2003.
52 Kamash 2016, 165.53 Kiernan 2020, 260.The transformation of Mithraea in the Late Roman period Romano-Celtic temple.Moreover, we cannot rule out the possibility that the smaller drystone structure also served a Mithraic function, possibly as an anteroom, given that there are no inscriptions or sculptures pertaining to the worship of a spring deity. 54 the London mithraeum, sometime in the early 4th c., the building experienced partial collapse. 55This may not have come as a surprise to the initiates, as it was not built on piles and was therefore susceptible to subsidence.The floor had already been re-laid several times in the half century since the mithraeum was built.As part of this restoration, the central aisle was raised to the same height as the benches, creating a level surface, and Mithraic sculpture fragments (alongside those from other deities such as Mercury, Serapis, and Minerva) were buried in the floor of the temple. 56Two altars were placed in front of the plinth on which the cult image had stood.It has been hypothesized these provided the base for a canopy under which a cult image sat.Wooden beams were also laid where the edges of the benches had once stood, suggesting that some form of tripartite division was maintained, albeit without the benches.This has been generally understood to mark the end of Mithraic activity in the structure, although the identity of the deity subsequently worshipped there has never been established conclusively.Bacchus has been suggested, given that a sculpture group consisting of the god with a retinue was found next to the north wall of the temple in the final floor phase.However, it has also been acknowledged that there is evidence to suggest a continued Mithraic use, including that the building contained an unusually high concentration of Camulodunum type 306 vessels both before and after the early-4th-c.structural alterations.The poor finish of these vessels may indicate they were intended for single use, most likely a particular ritual activity. 57ontinued Mithraic use may also be indicated by the fact that the relative number of chicken bones also remains high after these structural alterations, while a (now lost) ceramic vessel containing chicken bones was found in a pit dug into the last floor layer of the building before its final abandonment. 58 discussed previously, it cannot be assumed that the fragmentation and deposition of Mithraic sculptures is indicative of these objects ceasing to be of importance to a Mithraic 54 It is possible, of course, that the spring deity and Mithras were one and the same.The image of water pouring out of a rock after Mithras has shot an arrow into it is common in the Rhine area, and water, particularly from springs, played a major role in the cult.Many mithraea were erected over or adjacent to springs, with water from the spring channeled into a basin within the temple.The water from a spring would be reminiscent of the water flowing out of the rock pierced by Mithras, although the water could also represent Mithras himself, as he was born out of a rock (Clauss 2012, 72-75).In an inscription from Poetovio (Ptuj), Mithras is even referred to as Fonti Perenni (Vermaseren 1960, no.1533).The importance of the "watermiracle" in Germany is also evident from the initiation ceremony depicted on the vessel from Mainz, which shows the Father re-enacting this scene as part of the ritual, see Beck 2000.55 Shepherd 1998, 221, 227.On the mithraeum, see : Grimes 1968; Shepherd 1998; Gordon 2000;  Croxford 2003; Bricault and Roy 2021, 558-59.56   A relief depicting one of the Dioscuri was found in the mithraeum, although the exact context of its discovery has been lost, see Shepherd 1998, 182-83.It is tempting to compare this with the images flanking the entrance to the inner chamber at Hawarte and the possible amalgamation of the torchbearers with the Dioscuri.community.Moreover, it is possible that any Mithraic sculptures that were installed in mithraea during their last phases of use were removed upon the final destruction, as was the case with mithraea at Caernarvon, Colchester, Lambaesis, Lugo, and Orbe. 59In the case of London, several sculptures, including the small tauroctony dedicated by Ulpius Silvanus, were found in 1899.How they relate to the final phases of the mithraeum is unclear; it is possible that they were still on display in the temple. 60Incidentally, when the building finally went out of use sometime around the turn of the 5th c., several cattle skulls were deposited in the neighboring well, a strikingly coincidental act given the prominence of the bull in Mithraic iconography. 61That being said, the continuation of Mithraic activity in the building does not mean that it was never utilized for the worship of Bacchus either, for the Bacchus statue group was discovered on the final floor layer, laid decades after the early-4th-c.alterations. 62ong these atypical mithraea it is also worth including the mithraeum at Septeuil near Paris, which was installed within a former nymphaeum in the mid-4th c. 63 The mithraeum did contain two benches, but they did not align, with the bench to the left of the tauroctony considerably shorter (ca.3.5 m) than that on the right (ca. 4.6 m) to allow for the doorway in the northeast corner of the chamber, which also meant it was situated further west than its counterpart.A statue from the former nymphaeum, depicting the goddess of the spring, was installed in a niche in the south wall above the longer bench.That the statue did not originally belong here is evident from the fact its feet had to be removed to fit it into the niche. 64Again, the mithraeum produced evidence for feasting, much of which related to the consumption of chicken. 65nally, we may tentatively add a mithraeum constructed in 325 CE at Gimmeldingen in Germany.The structure is lost, but the inscription recording its precise date of construction refers to it as a fanum. 66This term has not been found in any other Mithraic inscription, for, as noted above, mithraea in Italy were generally referred to as spelaea (caves), while elsewhere they are usually called templa (temples). 67Fanum was a term often, although not exclusively, used to refer to Romano-Celtic temples, and its use here could suggest a structure in that style rather than a typical mithraeum.Incidentally, Mackwiller, with its mithraeum seemingly consisting of two concentric squares, lies only a day's travel from Gimmeldingen.Alongside the inscription, fragments of the tauroctony, reliefs depicting 62 Shepherd 1998, 91.Haynes (2008) has illustrated the potential for further research on relationship between Mithras and Bacchus by exploring parallels between the London mithraeum and the temple to Liber Pater at Apulum and warns against discussing such cults in binary terms.
The transformation of Mithraea in the Late Roman period the torchbearers, a relief of Mercury, and another relief possibly depicting Vulcan and Minerva were recovered. 68

Explaining the changes in Mithraic architecture
It should come as little surprise that a notable number of mithraea founded in the Late Roman Empire would break away from traditional bench-aisle-bench plan.Gordon argued that the founders of Mithraic communities were "inspired by a tradition… yet as leaders constantly on the look-out for new ideas and interpretations." 69There is no reason as to why these varied influences would not eventually spread to the architectural arrangement of mithraea, particularly given that, as far as we can tell, there was no written Mithraic doctrine that decreed all mithraea had to maintain the same plan.It is possible that these new "ideas and interpretations" included the incorporation of deities and iconography from non-Mithraic traditions that we find in association with atypical mithraea: possibly Zoroastrianism or Manicheanism at Hawarte; Isis in the Mithraeum of the Colored Marbles; the lararium outside the Via Giovanni Lanza 128 mithraeum; spring deities at Mackwiller and Septeuil; Mercury, Vulcan (?), and Minerva at Gimmeldingen; and the sculptures of various deities deposited in the London mithraeum. 70There has been a tendency to focus on the Mithraic iconography as the subject of veneration while treating these other images as ancillary, yet it is plausible that these objects were also the subject of ritual activity.In some cases, this might be obscured in the archaeological record due to similarities between Mithraic rituals and rituals conducted in honor of other deities.The worship of Mercury, for example, evidently involved the sacrifice and consumption of chickens, while a raven frequently appears alongside both Mithras and Mercury. 71That certain similarities between the gods led to them being worshipped simultaneously is evident from the various references to Mithras-Mercury in the epigraphic record. 72However, there are aspects of these deities that are notably different, such as the consumption of ovicaprids in relation to the worship of Mercury and the initiation rites involved in the cult of Mithras.It might have been a desire to accommodate such differences that led the leaders 68 Vermaseren 1960, nos.1314, 1316-22.69 Gordon 2017, 298.70   It is important to note that the presence of such images in mithraea is not uncommon and certainly not restricted to those with unusual plans: images of Mercury and Minerva are found in various "typical" mithraea (particularly in Germany) such as at Dieberg (Vermaseren 1960, no.1209), Nida (Vermaseren 1960, nos.1086, 1089), Stockstadt (Vermaseren 1960, no.1183), Merida (Vermaseren 1956, no.780), and the Foro Boario in Rome (Vermaseren 1956, no.441).On the frequency of images of Mercury in Mithraic contexts, see Hensen 1995.Older mithraea that remained in use into the Late Roman period have also produced images and altars relating to other deities, such as at Carrawburgh (Richmond et al. 1951, 31), Martigny (AE 1998, 872),  Santo Stefano Rotondo in Rome (Lissi-Carrona 1986, 39), and Trier (Walters 1974, 25).The various aristocrats in Rome who created their own Mithraic communities in the 4th c., while also adopting various priestly titles from other cults, might also be included in this evidence, see Griffith 2000. of some Mithraic communities to design their mithraea with an "atypical" plan. 73Ritual practices and their architectural settings are deeply bound together, and a change to one, even relatively minor, can have a notable effect on the other.Several studies have drawn attention to this both in antique contexts and more generally. 74 should also consider whether for at least some later Mithraic adherents, the interior design of their mithraeum was not of primary importance.It is difficult to believe that every Mithraic initiate, even the leaders, understood how a mithraeum might serve as a map of the cosmos or knew the possible meanings encoded in the tauroctony.Even if they did, they might not have considered this a necessity.Indeed, the very variability in the design of mithraea in the Late Roman Empire illustrates this.For many Mithraic initiates, the mithraeum may have primarily served as a clubhouse in which they built social relationships with their contemporaries and superiors.This is not to suggest that this is mutually exclusive with a genuine belief in Mithras, but rather that this belief was not the main reason for their attendance at a mithraeum.As we have seen, evidence for feasting, the most communal of Mithraic activities, remains consistent into the Late Roman period regardless of whether the mithraeum had a traditional plan or not.For some it might simply have not mattered whether the mithraeum had two, one, or zero benches, nor where the image of tauroctony was situated, so long as they went home having had their share of the feast and having reaffirmed the bonds among themselves.As Kamash has observed, "sensual memory practices on religious sites and particularly those associated with Mithraism can be seen as a powerful way of linking diverse individuals who were not anchored by a shared sense of place or history." 75Subsequently, such rituals developed and reinforced not only a sense of cult community, but also the community's attachment to the space in which they were performed.However, this did not necessarily mean that the space continued to be understood in quite the same way.Evidently in some cases there was a shift in how these spaces were perceived, as from the late 3rd c., mithraea started to appear that rejected the typical plan.
Wider socio-economic factors may also have played a role.It is perhaps no coincidence that atypical mithraea appeared at a time when the number of newly constructed mithraea (both standalone buildings and those installed in pre-existing spaces) entered into a decline (Figs.5-6). 76This decline may have been due to an increasing rarity of able and/or willing patrons, as was already the case for temples more generally by this period.77However, the 73 In some cases, this may also have been born out of necessity, as the general decline in temple construction and repair that is evident in the Roman West from the 3rd c. onward may have meant the worship of certain deities needed to be rehoused in other structures.On a decline in temple construction and repair: for Italy and North Africa, see number of mithraea installed in pre-existing structures/spaces does remain relatively more consistent into the 4th c. than their standalone counterparts (albeit with the number of those installed in pre-existing structures/spaces lower to begin with), which may have been due to the former being more economically viable.In any case, by choosing pre-existing spaces, these Mithraic communities were placing limitations on what could be achieved architecturally.At Septeuil, the design of the nymphaeum dictated that the benches had to be different lengths and unaligned.This might also be the reason why the Mithraeum of the Colored Marbles at Ostia contained just one bench, while at Lentia and Via Giovanni Lanza, the lack of benches might have been due to limited space within these rooms.That being said, the installation of several of these mithraea in caves (e.g., Epidaurum, Hawarte, and St Urban) might have resulted from a desire for a more "authentic" cave experience, rather than from financial restrictions.Indeed, meeting in an actual cave may have meant these adherents felt they could dispense with the typical mithraeum layout that was intended to replicate a cave.
In reality, there is no single explanation behind the emergence of atypical mithraea.Their very appearance is a testament to the variation that existed among Mithraic communities, each with its own aims and concerns.Some may have felt compelled to accommodate new forms of ritual practice, others may have been constrained by financial issues, and some may have felt it was enough to meet in an actual cave.Some of these factors are also not mutually exclusive, with the London mithraeum providing a possible example.The most probable explanation for the transformation of the inner chamber of this mithraeum in the early 4th c. is the initiates' frustration at constantly having to re-lay the nuanced image than analyzing all mithraea together, as I did in Walsh 2018a.While, as I have previously shown, the overall construction and repair of mithraea does enter into decline later than for other forms of temple, the decline in standalone mithraea from the mid-2nd c. illustrated here suggests broader socio-economic changes were having a stronger impact on these Mithraic communities than I originally believed.
floor and repair the damage caused by subsidence.But why did they not continue to replicate the original plan, particularly as the installation of wooden beams where the benches had been suggests an attempt to at least respect the typical mithraeum layout?Perhaps the presence of sculptures depicting various deities in the mithraeum had already had an impact on ritual practices and the new layout reflected such changes.Yet even if this were the case, it does not explain why, given the major structural issues with the mithraeum, the initiates did not just move elsewhere.Was it, as has been suggested above, now too difficult to afford the construction of a new mithraeum?Or did the communal memories of ritual practices conducted within the temple mean the worshippers were too attached to the mithraeum to abandon it, even if it meant changes to the internal plan?It might be that all of these possibilities contributed to the changes in the internal appearance of the mithraeum.While we can trace some broader trends to help us hypothesize, it must be borne in mind that every temple had its own particular founder, community, and environment, all of which contributed to its unique biography.

Conclusion
From the late 3rd c. onward a significant number of mithraea were constructed that did not conform to the typical plan so characteristic of earlier periods.This may be understood as a result of the increasing diversification of Mithras-worship.Just as those who founded Mithraic communities had adapted and or introduced new ideas from the earliest days of the cult, particularly with no texts to adhere to, it would seem only a matter of time before this had an impact on Mithraic architecture.Whether these changes in the layout of mithraea were the result of new ritual practices, or whether conversely, the architectural alterations forced changes to ritual practices, is unclear.Many of these atypical mithraea were found to contain evidence of the worship of other gods besides Mithras.While this is not unusual it might be that the desire to accommodate the worship of these deities in some capacity led to the creation of new mithraea with a layout that could facilitate this.In any case, it is highly unlikely that these new atypical mithraea contained a "star-map" and any rituals that were connected to this.Perhaps these mithraea were still understood to be caves, which some of them literally were.Many atypical mithraea continued to host feasts long into the 4th c., an element of continuity with the consumption patterns in earlier mithraea.
It is worth considering what this means for identifying mithraea, particularly those created in the later Roman Empire.Many atypical mithraea have produced iconography or inscriptions that indicate these spaces were used to worship Mithras.Had this not been the case, some certainly would have never been recognized as such.The mithraeum at Via Giovanni Lanza would appear to have been used as a storeroom and nothing else; the Mackwiller Mithraeum would have been identified as a small Romano-Celtic temple; and the Mithraeum of the Colored Marbles would perhaps be referred to as a room for Isiac feasts.Consequently, one wonders how many mithraea have been incorrectly identified due to their ground plans not adhering, at least at first glance, to the typical form.I have recently suggested that the late Roman "church" at Colchester may in fact have been a mithraeum: this case illustrates how important it is that small finds and depositional practices play a more prominent role in the identification of mithraea. 78A small temple at Marquise, near Calais, has been tentatively identified as a 4th-c.mithraeum due to the relative abundance of poultry and pig remains found in the building.Postholes within the building suggest a possible tripartite division, although it is unclear what form this took, and one might reasonably conclude that without the faunal remains, a Mithraic identification would not have been put forward. 79It is quite likely that Mithraic communities were more common in the 4th, and perhaps 5th, centuries than has traditionally been thought, but their archaeological traces need to be detected via more nuanced means.

71
King 2005.The aforementioned chapter by Haynes (see note 62) also draws attention to the ritual use of Camulodunum type 306 vessels in both Mithraic and non-Mithraic contexts.This includes material from the "church" at Butt Road, Colchester, although I have reinterpreted this structure as a mithraeum, see Walsh 2018b.