A rock-cut tomb of the Mongol period in the Ilkhanid capital of Maraghe

Abstract

By the time of Kublai’s death, the Mongol Empire had fractured into four separate khanates or empires including the Golden Horde [Kipchak] in the northwest, the Chagatai Khanate in Central Asia, the Ilkhanate, based in modern-day Iran, and the Yuan Dynasty in the east. Although the burial architecture of the Yuan Dynasty and their funeral ritual have been the foci of the archaeologist, there have been no convincing conclusions about the other branches of the Mongol Empire. According to historical literature, all of the Ilkhanid Khans before Ghazan (1271–1304 CE) were buried in unknown places after their death. Archaeologically, not only we have no clue to trace the royal tombs of the pre-Ghazan period, even the identifiable tombs of the royal family members of the Ilkhanid Iran and high-ranking Mongol nobles have not discovered. Taking Maraghe, in Northwest Iran, as the first Ilkhanid capital (1256–1265 CE), the aim of this paper is to study the archaeological remains of the enigmatic rock-cut complex in the village of Varōy [Varjavy] to provide a more detailed description of the current remains. The results show that, while reassessment of the possible functions regarding this site suggest and outright contradict to the traditional views of scholars as a mithraeum, the architectural layout of this building is deeply intertwined with Mongol funeral sites and has much to tell about the relationship between rock-cut complexes and Ilkhanid cemeteries.

1 Introduction

Kuh i-Sahand (Sahand Volcano) which covers an area of about 2800 km² (Mehdipour Ghazi et al. 2012) is particularly rich in rock-cut architecture compared to the other parts of Iran and needs further investigations.¹ The softness of the volcanic tuff layer makes rock-cut architecture possible around this mountain, and never becoming very old.² The major rock-cut sites discovered around Sahand volcano are Kandovan, Qadamgah, Girkh Kuhul (Sattarnejad 2017), Kuhullu Machid, Pir Hashem (Sattarnejad 2019), Gowijeh Qaleh (Naseri et al. 2015), and Kahnamu (Shekari Niri 2000). A challenging problem that arises in studying this monument is that unlike the adjacent archaeological sites, this building lacks a vast corpus of textual sources, as is often the case for rock-cut sites. In the history of Zoroastrianism³ or Mazdayasna, western Iran has been thought of as the heart of religious activity under the Median emperors (Herodotus 1920, 203). It is believed that after the unification of the Median and Achaemenid dynasties in 550 BCE, temples were constructed throughout the empire to highlight the role of Mithra as a Zoroastrian divinity (Malandra 1983, 88). One major theoretical issue that has dominated the literature on rock-cut architecture of northwestern Iran for many years concerns the Roman mithraeums, which are mainly underground structures (Clauss 1990, 70). As a result, several studies have sought to conclude a mithraeum function⁴ for almost all troglodyte sites around Sahand Volcano Varjavand 1976; Azad 2005; Labbaf 2012; Mirfattah and Shekari Niri 1996; Shojadel 2000a, b; Soltani 2018). Like other rock-cut sites in Northwest Iran, we cannot clearly identify the time of construction, function, and abandonment of the underground rock-cut complex at Maraghe due to a lack of precise chronological systems and insufficient and fragmentary excavation data (Fig. 1). Considering these challenges, it is worthwhile to discuss the remaining materials as the only reliable evidence. This building is important for two reasons: First, it points to non-Muslim architectural patronages at a very high level in Northwest Iran and offers evidence that is not preserved in literary or other sources (Moradi 2020, 2022). Second, it bears witness to the practice of reusing non-Muslim buildings in the Islamic context.

Fig. 1

Location of site and Sahand Volcano in Northwest Iran (Encyclopedia Britannica, Inc)

2 Literature review

Although the importance of Maraghe for the architectural history of the Islamic period has been acknowledged,⁵ explanations for the appearance of rock-cut sites in this region is not satisfying. Regarding the underground rock-cut complex at Maraghe, there are a few theories concerning its origins and purpose, none of which can be supported by hard evidence. Varjavand (1972) was among the pioneers who conducted a series of archaeological excavations to study this site. In his report, he proposed a definite Parthian (247 BC-224 AD) settlement at the site, which led to an Imāmzādeh function in the Ilkhanid period (13th century).⁶ With this presumption that the architectural layout of this monument has no resemblance to the common religious practices in Iran, the function of this building was surmised to be a mithraeum (Varjavand 1976: 14). This hypothesis produces a deep chronological gap from 247 BC to 1256 CE, while archaeological evidence does not go beyond the 13th century. Ball (1979) provides a more accurate overview by focusing on the inscriptions and decorated stonework in the complex and the adjacent cemetery. With the presumption that Abaqa Khan (reg. 1295–1304 CE) had visited a butkhana⁷ near Maraghe (Rashīd al-Dīn 1174: 982), Ball suggests that this building is one of those butkhanas configured as an Islamic building when Ghazan Khan (reg. 1295–1304 CE) converted to Islam. Shekari Niri (2006, 2020) holds the view that the ground plan of this building reflects the figural representation of Mithra that appears in Roman mithraeums. Without considering construction phases, he states that the whole site dates back to a pre-Islamic construction methodology, specifically the Parthian period. Considering the fact that Kashmir was an intermediary between Persia and Central Asia in a cultural sense, a point that is reflected in Ball’s contribution, Azad (2010a) categorizes this building as an Ilkhanid structure with probable Buddhist characteristics.

The function of this building has recently been reviewed by the author (Moradi and Omrani 2020; Omrani and Moradi 2021) demonstrating the fact that the attribution of a mithraeum and a Buddhist temple is far beyond the archaeological evidence and the whole complex must be studied within the context of the funeral sites.

3 Methodology

The research method for this paper starts from much broader aspects of architectural details and decorations by emphasizing the field works to studying the characteristics of this underground rock-cut architecture in order to highlight its Ilkhanid aspects. This also includes the examination of previous archaeological investigations, which can be usefully compared with the data provided by the research. Direct analysis of evidence from the interior surfaces with particular reference to architectural developments was used to understand the dynamics of major changes in this monument. This makes it possible to propose a typology and attribution regarding this building’s identity and will have broader implications for the question of the probable function of this curious monument.

To offer a function that fits the architectural remains, I would like to study this rock-cut site in relation to other sites within the region and period under study since developing new hypotheses irrespective of architectural parallels leads to misleading attributions. Looking for similarities promises a more flexible hypothesis, so I discuss these sites, not with respect to how they differ from each other but how they are similar in form and function, as a way to identify essential links.

4 Description

The underground rock-cut complex of Maraghe is an irregularly shaped entity located on the southern slope of Sahand Volcano, some 6 km southeast of Maraghe and arranged along a north-south axis. There are various ways of spelling and pronouncing the name of the village in which the site is located. The most popular, and the one by which it is generally known in the local Azerbaijani literature is “Varōy”.⁸ Without pretending to any etymological correctness,⁹ or to any hypothesis regarding its origin or history, Varōy indicates a close linguistic link to the name of the village of Varāy (Viyār) which is located near the late Ilkhanid capital of Sultaniyya, and is known for the impressive rock-cut architecture called Dash Kasan.¹⁰

Approaching the complex from the north, a stone stairway leads down to a rectangular hall, which is the spine organizing the interior of the surviving sections of the complex. The present staircase might belong to a newer phase of construction since the builders of this monument could easily create rock-cut stairs wherever they wanted.

A giant dead trunk of an old tree was once located outside, in the main axis of this building.¹¹ It is unknown whether this underground architecture was originally a natural cave that was improved and furnished or whether it has been completely hewn from the bedrock. However, evidence of expansion shows that the complex was not planned as a single unified project, but rather was adapted to pre-existing structures on the site. The width of the entrance space ranges from 6.20 to 7.80 m, while a big portion of its ceiling has crumbled away. Therefore, the final form of the entrance cannot be ascertained. The length of the corridor reaches 16.25 m before leading to the rectangular hall (Fig. 2A). The door frame is well preserved and can only be seen when standing in the vestibule. On the left-hand side when facing the entrance, there is a rectangular hall divided into four compartments, and supported by an octagonal pillar in the center. Its total length is 9.81 m and its width is 8.90 m and believed to be a khānqāh (Shekari Niri 2006) (Fig. 2H). The central rectangular chamber that covers approximately 19m² has access to all the spaces (Fig. 2B). Although there are no remains of a screen in the main entrance (the passage between spaces A and B), it has sockets for a probable wooden door. It is not clear to what extent the deformation of the original roof can be attributed to human intervention or erosional processes. The thickness of the tuff layer from the ground level to the ceiling in all parts ranges between 0.7 and 1.30 m. Such thickness eliminates any potential for water leakage caused by possible heavy rainfall.

Fig. 2

Overall plan (up) and section (down) of the underground rock-cut building at Maraghe. A: Corridor; B: rectangular hall; C: main circular room; D and E: lateral rooms; H: hall of four compartments; J: underground cells

Remains of a partially vanished Mihrab and a horizontal text band in the Naskh calligraphy are evident on the western wall of the rectangular hall (Fig. 3). The inscription is from the fourteenth verse of Surat al-Fath.¹² The Ayah inscription that runs in one line on the western wall of the rectangular hall continues to the circular room at the same height. This is the only evidence pointing to the fact that spaces B and C surely existed during the decoration process. The other important feature is that the main part of the inscription in space B has been destroyed by adding a portal to create space E. It can be concluded that the inscription band has been damaged for further expansions and space E is definitely a later addition. Although door sockets are seen on either side of the entrance of this room, the narrow width of the entrance of this space (0.68 m) makes it unusual for constant daily use. Surprisingly enough, small niches for oil lamps are absent all around the rectangular hall and had probably been replaced by a fairly large niche under the inscription. Since the main instruments of interior lighting in medieval ages were oil lamps and candles (Hillenbrand 2015), it is quite possible that this niche was used as a place to set lamps. In the Islamic faith it is believed that on the Day of Judgment, the believers will get the intercession from the Prophet of Allah (Arifin 2019) and lighting a candle can ease the connection with the buried person inside the Imāmzādeh for his heavenly assistance. The existence of a votive candle area has remained a favored element of Islamic architecture and even today many Imāmzādeh functions have a version of this design.

Fig. 3

Western wall of the rectangular hall in which the inscription has been damaged to create space E. Sockets to hold a wooden door are seen at the right corner of the portal

The framing decoration of the entrance, which is much damaged but recognizable, is adorned with two bands of interlaces that frame a palmette pattern. Unlike other interior spaces that show no signs of decoration, the main circular room (space C) represents a very carefully finished interior with a muqarnas ring and is about 95 cm higher than the floor ground of the rectangular hall (Fig. 4). This room has a similar plan to the lateral circular rooms but its dimensions exceed them. The Arabic inscription of the rectangular hall was planned to run beneath the muqarnas ring, but it was covered over by a thin layer of mortar (Razani and Hamzavi 2018) before its execution. Obviously, the execution of the other circular rooms is inferior to that of space C, not in richness and quality of ornament, but in quality. The evidence of red pigments in the interior surface of space C, under the muqarnas ring, can be affirmed. Space C is the only two-storied room at this monument and a staircase in the entrance leads into two underground cells with a curved ceiling (Fig. 4-J). A narrow tunnel in the internal wall of space C leads into two irregular consecutive rooms that meet the main hall. The triangular arch of this channel resembles the arches of the Rasad-Khana¹³ caves on a smaller scale. However, the small dimensions of this unit are intriguing when considered as a channel of aeration. The square chamber (1 m × 1.36 m) at the southern side of this room does not appear to have been quite finished and represents a later addition as well, since this space is unfinished and the working marks in its interiors is very rough.

Fig. 4

Interior view of space C and the entrance of the cells

5 The reason why the underground rock-cut site at Maraghe was not a mithraeum

The extensive debate about the function of this complex in previous studies invites a review of the arguments that have led scholars to consider this structure as a mithraeum. Cumont (1896) argued that Roman Mithraism is an offshoot of the Persian god of Mithra that is mentioned in the Avesta, the collection of religious texts of Zoroastrianism.¹⁴ From this point of view, the appearance of the cult of Mithra should be traced both to the Parthian (247 BC- 224 AD) and Sassanid Empires (224–651 AD), during which the surviving text of the Avesta and the recension were produced (Humbach 1991: 45). Although Dura Europos in Syria was a tributary of these empires from 165 AD to 256 AD (Dirven 1999: 87), archaeological evidence related to mithraeums extended merely from Britain to Dura Europos. Otherwise, present-day Iran appears to have been less influenced by either Roman or Parthian Mithraism, and there is no example in Iranian architecture that satisfies the general requirements of a mithraeum (Clauss 2000: 16). On the other hand, we should ask: Does the architectural layout of this building fit into the course of developments of mithraeum architecture or not? Unlike the typical layout of a mithraeum with an east-west orientation, this building is set toward the south. From an architectural point of view, a narthex, nave, apse, side aisles tauroctony reliefs¹⁵ (Shepherd 1998: 227), and additional ritual elements that may represent the seven grades of the cult¹⁶ and are known to have existed in many other mithraeums are absent in the underground rock-cut architecture of Maraghe. The exact analogy of the common architectural layout of a mithraeum is also found even in the rock-cut mithraeums like Hawarte (Gawlikowski et al. 2011), which would suggest the fact that the function of the rock-cut complex at Maraghe as a mithraeum is debatable.

Moreover, evidence of a religious feast, mainly consisting of sheep, pigs, cattle, and chickens is reported from the archaeological layers of almost every mithraeum (Shepherd 1998: 227).¹⁷ Hence, remains of furnace and ashes would be inevitable for cooking the meal within the building (Vermaseren 1956: 41). Although a charcoal layer is expected to be a significant feature in archaeological excavations of mithraeums, as was the case in the mithraeum at Kempraten, Switzerland (Russo et al. 2018), there is no conclusive evidence of any meal remains nor scorched areas on the floor of this building at any part other than around the lamp niches. Furthermore, water plays an important role in the construction of a mithraeum for ritual purposes, so most of the mithraeums are located near water sources such as Mackwiller, Poetovio, Housesteads, and Gallia Belgica (Irby-Massie 1999: 78). In contrast, here, there is no water source or hydraulic system to bring water into the building. Additionally, tombstones in such a number assert continuous funeral activity which is unprecedented in connection with a mithraeum.

6 Discussion

In the Ilkhanate period (1256–1335 CE)¹⁸, this building was a focal point for a cemetery and Shekari Niri (2006) has recorded these tombstones based on their inscriptions (Fig. 5). Therefore, one might consider that the location of this underground rock-cut architecture was a possible impetus for the distribution of tombstones across the site. The other rock-cut sites with a similar combination of a great number of tombstones that are dated to the Ilkhanid era by inscriptions are Qadamgah in East Azerbaijan province (Karang 1972: 490) and Abazar in Ardabil province (Mirfattah and Shekari Niri 1996), both in northwestern Iran. Comparisons to contemporaneous sites reveal the fact that the emergence of these sites is inherently linked to the Ilkhanates, since definite evidence of a pre-Ilkhanid box-typed tombstone has never been reported or documented in this region. Although this hypothesis needs much scrutiny, there are some immediately dependable conclusions that the advent of an Ilkhanid cemetery and a rock-cut site may propose the idea that they are relevant to the Ilkhanid heritage. However, further studies, which take this hypothesis into account, will need to be undertaken. In addition to the geological capabilities of Sahand Volcano, we must point out that this volcano occupies a prominent place in the history of Mongols, and Ilkhans had ample reason to choose this terrain. In reviewing the literature, this region is known worldwide for the four regional historic contexts including Maraghe, the first capital of Ilkhans (1256–1265 CE) in the south; Ojan, the summer camp of the Ilkhanid royal family in the east (Velayati et al. 2015); Tabriz, the most important city of the thirteen-century Iran in the north, and the Shahi Island in the west where the mysterious tomb of Hulagu Khan (1218–1265 CE), who laid the foundations of the Ilkhanates, is located (Rashīd al-Dīn 1994: 1022). If we consider the fact that the major mythical events of The Secret History of the Mongols¹⁹ takes place in a mountainous landscape, it can be concluded that Sahand Volcano and the distinct quadruplet regions around it could guarantee the perpetuity of this terrain as an extremely crucial zone for the Ilkhanates, where the Ilkhanid empire had risen (1256 CE) and declined (1335 CE).²⁰ This building was also supposed to be the tomb of the Mullah Māsum e-Maraghei who was believed to be a famous member of the clergy in this region in the 13th century and was considered to be an Imāmzādeh, a descendant of a Shi’i Imam (Mokhlesi 1992: 168) but there is no historic text confirming this. Interestingly, even today, groups of people gather here every year to attend the Mourning of Muharram over Imam Hussein’s death.²¹

Fig. 5

Topographic plan of the site including the underground structure (ground plan in red) and remains of the Ilkhanid cemetery

The pronunciation of the nearby village in Azerbaijani language differs somewhat from those of standard Persian and “Varjavy” is the Persianized form of the “Varōy”. In his " Dīwān Lughāt al-Turk” (compiled in 1072 CE), Mahmud i-Kashgari refers to the word “Vara” that is phonetically and graphically so close to the name of the rock-cut sites both in Maraghe and Dash Kasan with special reference to the myth of the separation of sky and earth (Kashgari 1954: 211) 9Fig. 6). This expression might be reasonably traced back to the religious believes regarding death and it is quite possible that “Varōy” and “Varay” are semantically similar words in the Turkic language that have been preserved throughout history and transferred to later periods.²²

Fig. 6

Interior view of the rectangular hall to the south

The emphasis on Quranic text around the carved Mihrab may serve the specific purpose of displaying the religious function of the building, however, it is essential to evaluate some problematic points regarding the carved Mihrab on the western wall (Fig. 7). Since west is not the direction that Muslims should face when praying, this orientation might refer to the Mongol shamanism where the west was the home of the good male heavenly sprits (Stewart 1997), or it can be attributed to the Chinese Buddhism interest, in which the west represents movement toward the Buddha or enlightenment (Lagerwey 2009: 108). Regarding the former hypothesis, the non-Islamic trans-continental influence on Ilkhanid Iran is inevitable since Mongol rule in Iran fostered the direct exchange of ideas and practices between diverse cultures and religions (Azad 2010b).²³ In the regional architecture, the appearance of a rock-cut Mihrab on the western wall foreshadow the two similar Mihrabs on the western wall of Dash Kasan, also carved in stone, that bears traces of the patronage of the Ilkhanid empire (Scarcia 1975).

Fig. 7

Details of stone-carved Mihrab on the western wall of the rectangular hall

The same interest to create stone-carved muqarnas can be seen on the countless Ilkhanid tombstones in northwest Iran. Since there were no earlier examples of stone carved Mihrabs in Iranian architecture, I suggest that such models for architecture and its decoration existed exclusively in thirteen-century northwest Iran, pointing to a local tradition.²⁴ In any case, the appearance of a Mihrab on the western wall confirms the fact that this building have never been used as a mosque.

Based on the calligraphy inscription, it appears certain that the building was in use during the Mongol domination of Iran (1256–1335 CE). The greater depth (about 0.7 cm) and size of carvings in the interior space argues for a greater sophistication in decorative conception compared to the floral patterns of the main entrance and demands much greater expertise. This is in outright contrast with the maximum depth of the slightly lighter carved foliated patterns of the main entrance (0.3 cm), but it is not necessarily related to the importance of this hall compared to the other spaces, since the outer carving could be easily weathered due to its insecure position. Moreover, the inscription reveals a strong local influence that resemble the Ilkhanid tombstones in northwest Iran, suggesting that this style had become part of a standard repertoire and implying that a local workforce participated in its creation, even if some of the workers were brought in from elsewhere (Fig. 8). Even though the font of the inscription in gravestones is closely intertwined and compressed within a defined space in a geometric frame, their Naskh scripts proved their calligraphers’ artistic links to the oblong inscription band in this rock-cut building.

Fig. 8

Detailed view of inscription band in the underground rock-cut architecture of Maraghe (A) and Ilkhanid tombstones of Alanjag (B), Rab-e-Rashidi (C and D), and Kuhul (E)

In the pictorial documentation before the 1970s, the rectangular hall was covered with a makeshift metal dome resting upon a brick structure. Several fragments of Ilkhanid tombstones were also reused as building material in the construction of the base of this structure. This raises a further question concerning the original form of the probable domes in this area, since Iranian architecture generally lacks rectangular domed chamber rooms. Hence, the question of whether or not the original ceiling was covered with a masonry dome would be a source of academic contention. It is also possible that a convex flat ceiling cut down into the living rock once covered this space, but we have no evidence to suggest that this was the only case. In any event, it is hard to believe these openings were not covered since seasonal flows of heavy rains could easily access the building and cause damage. As remains of the later interventions in room D attest, one structural solution was to provide a proper drum between the circular base of the dome and these openings to level up the dome from the exterior ground level. However, since the maximum diameter of these openings do not exceed 3 m, the exact geometry and the structure of them was not a serious concern and did not require heavy continuous masonry shields to insure its structural integrity. Then, the final geometry of the probable domes was ruled by the individual of the mason rather than specific rules without having to deal with serious structural challenges.

The fact that the decoration of this building mainly depends upon foliate patterns is indeed remarkable. In this respect it contrasts strongly with the decoration of Ilkhanid masonry buildings in northwest Iran, where the taste for geometric decoration is very pronounced (Moradi and Omrani 2014). In visual terms, the details of the decoration of the entrance door in this rock-cut complex call to mind the Iranian stucco work of the twelfth to fourteenth century. The connections between the motifs are indeed compelling, especially with regard to the shape of the leaves of the palmettes, but they always appear in stucco and never in stone. The only equivalent stone-carved decorations in the closest neighboring society are traceable to Armenia, seen on mausoleums, khachkars and churches. This could be explained as the result of direct involvement of the Armenian masters or even their ideas, however, we may easily imagine that new results like these arose from the collaborations between local stone-carving experts and immigrants who brought their designs with them. Palmette pattern in Geghard (1215), Gandsassar monastery (1216–1238) and Hovhannavank monastery (1215), near Yerevan (Mahé and Kazaryan 2015: 21), suggest that, at least some technical influences in terms of stone-carved decoration can be found between Armenia and northwest Iran. The primary sources from the early 12th century clearly exhibited the relationship between the Ilkhanid Iran and the Armenians as their vassals (Prezbindowski 2012).²⁵ Then, it is convincible that due to the economic and political changes the Armenian inhabitants were present in the Ilkhanid society or the Ilkhanid patrons invited masters from this region into the Ilkhanid capital of Maraghe and used them in their architectural projects. If we accept the idea that Armenian masters may have been involved in the decoration, as they had the centuries of expertise to undertake such work, the stone-carved decoration of the entrance in the Maraghe troglodyte could belong to the time frame between 1236 until around 1335 during which Armenia was tributary to the Mongol Empire (Ibid) (Fig. 9).²⁶

Fig. 9

Stone-carved decoration in the dihedral niche on the southern exterior wall, monastery of Hovhannavank (left), and the rock-cut complex of Maraghe (right)

Like the decorations of the main portal, the existence of stone-carved muqarnas might point to the involvement of craftsmen familiar with the stone material (Fig. 10).²⁷ In the nearby geographical boundaries, stone-carved muqarnas appears in Dash Kasan and the portal of Saint Stepanos Monastery, some 15 km northwest of the city of Julfa -in Azerbaijan- as well. Unlike the abovementioned sites, the muqarnas of the large circular room is of a very special type and presents a homogenous entity hewn out of rock and provides a close technical comparison to Armenian architecture. Its execution is something quite new among Iranian architecture since in the Persian style, muqarnas are typically brick or ceramic tiles embedded into the mortar layer. In the gavit of the monastery of Geghard (1215–1225), as well as the close by cave church at the same site built around 1283, a complex muqarnas vault cut in a dark volcanic tuff layer that covers the central square (Ghazarian and (Ousterhout 2001: 144), provide excellent indicators of Armenian architectural inventiveness. Although stone-carved muqrnas fits in well with the tendency towards local styles within Anatolia from the second half of the 13th century onward (Blessing 2014: 129),²⁸ we know that Armenian masons are also documented among the Seljuk Anatolia, participating in construction projects (Blessing 2014: 201). Therefore, it is reasonable to conclude that the strong political coalition between the Mongols and the Armenians must have facilitated a common architectural practice.²⁹ Since monolithic muqarnas is unique for the early thirteen century architecture in Armenia, the appearance of that in this rock-cut complex suggests that the masters may have been influenced by the Armenian architectural heritage of this age.³⁰

Fig. 10

The composition of muqarnas decoration in the rock-cut complex of Maraghe (up) and the monastery of Geghard in Armenia (down)

It is evident that spaces D and E are the cheapest and least laborious spaces by which buildings of this type could be erected. Whether spaces D and E were cut simultaneously or not is difficult to determine. From the similar dimensions of their ground plan (diameter 2.70 m) and maximum height (approximately 3.30 m), one might conclude they were, and there is certainly nothing to contradict the supposition. From this point of view, specific sensibility to space C inspired a later generation of architects to produce a similar plan to expand the interior space of the whole complex.

The existence of underground cells in space C is similar in almost every respect to the lower story of the neighborhood Ilkhanid rock-cut architecture of Rasad-Khana caves as well. The fact that these two buildings present such remarkable parallelism must surely be more than coincidence. In its current situation, this building is believed to be a tomb of a sacred person. Then, these rooms would have been used to hold dead bodies, combining a crypt and the upper room as is the case in many Islamic tombs.

I suggest that the unfinished interior walls of the burial chambers under space C belong to the expansion phase when a wooden lattice structure (Zarih) was installed on the sacred case (Sandoug) (Beigbabapour 2009: 179), a phenomenon that gives the appearance of an active burial place for this monument. It suggests that the entrance of this room is potentially attractive to excavate the entry of the underground cells, however, the skill used to construct these chambers is in an outright contradiction with the well-polished upper story. On the other hand, the selection of this point might create limited access to the sacred area (space C), which is a common characteristic of almost every shrine as a way to raise the status of the deceased (Goudarzi 2017). These cells are spacious enough (ca. 2 × 4.30 m) for both coffin burial and placing dead bodies and from their partially symmetrical plan it is possible to conclude that they have been cut simultaneously (Fig. 11). However, one might suggest that a body would either be contracted in some way, or even that they were used for the deposition of bones moved from other graves as a secondary burial or even the cremated ashes of corps. The latter would be reasonable since these spaces could not be filled by soil after a fully extended burial.³¹

Fig. 11

Ground plan of the second floor under space C

The orientation of the tombstones in the adjacent cemetery provides an important opportunity to advance our understanding regarding this building. Here, tombstones were aligned on an east-western axis which is the standard characteristic of Muslim communities in burying the deceased with their head facing Mecca. Graves of this type with a box-shaped tombstone are not limited to this site but also distributed all around the 13th and 14th century northwest Iran. In 2008, Jian Wei excavated the Yike Shu Cemetery, some 12 km to the northwest of the Shangdu City, which belongs to the Yuan Dynasty (Wei 2008). In the following year, Ulambayar Erdenebat studied the old Mongolian graves between 11th to 17th century in Mongolia (Erdenebat 2009). The graves recorded by these studies are of importance here, as they portray striking similarities with the orientation of the burial cells under space C. Unlike the surrounding tombstones, the alignment of these cells represents a north-south axis that connected with the Mongolian cult in which particular attention is given to the sunrise with the head facing east (Halbertsma 2005: 161). Additional evidence from the prehistoric Mongol graves in Alag Tolgoi and Khanan cemeteries in the northern Mongolia, all of which represent a field of Mongolian burials of the imperial period (Nelson et al. 2011 :216), indicates the fact that the north-south orientation was the standard norm of inhumation from the Early Iron Ages (ca. 800 − 400 BC) in Mongolia. Therefore, it is reasonable to conclude that people who had designed and cut these burial cells or the occupants followed the Mongolian tradition.

With the presumption that space C once was coated by a white lime mortar (Razani and Hamzavi 2018), it is likely that the whole interior space once was covered with a red pigment or even decorated with red geometric patterns. While the latter composition is the common characteristic of the rock-cut architecture in the neighborhood region of Göreme,³² the Cappadocia region of central Turkey (Carpiceci et al. 2007: 229), the former adhere to the local architectural heritage in northwest Iran and finds parallels in Behestan in Mahneshan County in Zanjan Province, probably built in the 13th century, Jiraghil, a rock-cut village in Yengejeh Rural District, East Azerbaijan province. In these sites, after preparing the wall surface and plastering it with clay, gypsum, or lime, the whole surface was covered with a red pigment to give an eye-catching colored appearance.

Unlike neighboring rock-cut complexes of Qadamgah in Azarshahr and Kuhul in Marand, where fire was set against the tuff layer to heat the surface, which was then doused with water, causing the tuff layer to fracture by thermal shock, there has been no evidence of the fire-setting technique in this complex. It is likely that because of the softness of the tuff layer, stone masters were able to cut the spaces using primitive quarrying instruments such as hammers and chisels. We do not know from which area masters who were involved in cutting this complex came from, it is quite possible that masters (Muqanni) who were specialists in Kariz³³ engineering were participating in these projects but convincing written sources are still missing.³⁴

The building chronology could be divided into two phases. During the first phase of construction the original building was built on a north-to-south axis hewn into the rock. This included an ante-chamber, rectangular hall, and circular room.³⁵ Apparently, at this level, and for an unknown reason, the idea of carving inscriptions along the internal spaces was replaced by adding a thin layer of mortar on the surfaces. This plan continued to be used through the second phase, and the building was extended by excavating further spaces on either side of the rectangular hall. These expansions could have occurred both in the Ilkhanid era and later periods, but the former scenario is more likely.

Ball’s (1979) hypothesis of a butkhana and a Buddhist temple deserves comment. I quote Abū Yaḥyā Zakarīyāʾ b. Moḥammad Qazvīnī’s account (13th century) of a hewn cave in Maraghe in his Āṯār al-belād wa aḵbār al-ʿebād (Monuments of the lands and historical traditions about their peoples): “In the suburb of Marāgha there is a cave in which many rooms have been cut off. No one dare to enter as the hanging sculpture at the end of the cave is believed to be a talisman protecting a treasure” (Qazvīnī 1994: 642).

A religious symbol in a cave is the only evidence that could have inspired Ball to mark this place as a butkhana, since Ilkhanid Iran had significant contact with Buddhism through neighboring Kashmir, where Buddhism and its related science were very active (Tlalim 2016: 4). Although we might come to the conclusion that there were exchanges of architectural knowledge and style between these two regions, it is possible to think of anywhere in the suburb of Maraghe where such a cave could be. Then, Qazvīnī’s description does not necessarily act in favor of the connection with this monument. In other words, the existence of a large number of rock-cut caves around Maraghe, like Girkh Kuhul, Kuhullu Maschid, and Gowijeh Qaleh, that could fit the description casts doubt over this conclusion.

If we accept Ball’s (1979) idea of a Buddhist temple for this building, the peripheral cells, some 20 m to the west, could be marked as residential areas for monks and pilgrims, but again, the evidence from archaeological investigations indicates no remains of heating units or even a centimeter of ash layer in these probable lodging spaces. This makes the Buddhist temple conclusion problematic for a number of reasons: how is it possible to pass the freezing winters of northwestern Iran in such cells without any heating facilities? We must bear in mind that if the building was not resided in constantly but rather was subject to several migrating camps during the year, this would overshadow the holiness of the place. The other objection is that, while the main entrances of Buddhist monasteries are mainly located on the southern sides of the monastic complexes (He 2013), in this building, it is located on the north. Moreover, Rashīd al-Dīn compares Ghazan Khan with Prophet Ibrahim who ordered his governors to be doing all they can to eradicate any sign of Buddhist temples in the Ilkhanid realm to herald in the age of the monotheistic faith of Islam (Rashīd al-Dīn 1994: 1356).³⁶ Finally, while there are countless rock-cut Buddhist temples all around the world (Mason 2017), no similar underground Buddhist temple has been reported anywhere else. Therefore, we are permitted to believe that this building does not meet the criteria of a Buddhist temple.

The ground plan of the rock-cut complex of Maraghe is extremely rare in comparison to the architecture in other parts of Iran. As mentioned earlier, stylistic features of this building make it close to the rock-cut buildings of the same age, like Qadamgah³⁷ in Azarshahr, and Abazar (Shekari Niri 1996), where an attribution to the Ilkhanid era is likely based on nearby tombstones. In the Islamic faith it is believed that delivering and sharing the social dignity of a notable deceased person is feasible by having a grave in the vicinity of his or her tomb.³⁸ Although historical account exists to explain the development of such funeral enthusiasm from Ilkhanid period, fragmentary evidence suggests that this tradition has always existed in Islamic culture (Parsapajouh and Terrier 2019), thus it is reasonable to conclude that the surrounding graves were chosen to guarantee the salvation and intercession of the deceased in the resurrection.

It is believed that the eastern square hall served as a Khāneqāh, a center of a Sufi brotherhood (Shekari Niri 2020) (Fig. 12). Although the architectural layout of a Khāneqāh is usually simple and undemanding, including a large main room where the Dhikr ceremonies and the communal ritual prayer take place, related historical accounts do not exist to confirm this hypothesis about this building.

Fig. 12

Ground plan of the eastern hall (space H) (right), and Details of rock-cut niches in the eastern hall (left)

Even if we consider this hall as a Khāneqāh, it would emphasize the function of the main part as a burial place since in Ilkhanid Iran, Khāneqāhs like those in Ghazaniya in Tabriz (Rashīd al-Dīn 1997: 1321), Natanz in Isfahan province, and Bastam in Shahroud city of Semnan province are in connection with the burial tomb of the Khan or Sheik (the spiritual guide). Another angle to the function of this space looks into the existence of a set of four niches with pointed arches on four sides of the hexagonal pillar as well as the interior walls (Fig. 12-left). Depending on the size (40 × 25 × 3 cm) and composition of these elements, it is tempting to consider this room as a pigeon house which finds similar rock-cut parallels in the neighboring Ilkhanid villages of Kuhul in Marand, Jiraghil and Majarshin in Azarshahr, and Khiljan in Sard-e Sahra Rural District, in the Central District of Tabriz County. Although the primacy of relationship between birds and men in Iran is ancient (Goodell 1979), it is only in the Ilkhanid era that the pigeon appears in Persian literature history.

According to Rashīd al-Dīn, the pigeon was held to be a sacred bird and Ghazan has ordered to provide them with their basic needs like food, water, shelter, and a safe place nearby the mausoleums and shrines to raise a family in the spring (Rashīd al-Dīn 1997: 345). However, it should be mentioned that the pigeon differs from poultry in that it is strictly a flying bird and likes to mount higher objects (Dietz 1922: 5). For this reason, the pigeon house should be elevated as much as possible which is the case in all of the rock-cut pigeon houses in northwest Iran. Then, hypotheses regarding its function is still waiting for further comparison and confirmation.

7 Probable function

Referring to field survey investigations and on-site visual inspections, the initial construction phase of this monument comprises a passageway leading to a rectangular chamber in connection with a rear round chamber, all placed along an axial line (Fig. 13). As to the geographical area, this articulation finds similar parallels in Bashan’s tombs (926 AD) in Inner Mongolia (Hung 2013) (Fig. 13I and J). Although it is not easy to conclude the exact origin of this plan in this region, it is interesting that a fairly similar plan was an important feature of the Northern Wei royal tombs that were built some 400 year earlier than Bashan’s tombs (Ibid). Tomb No. M67 at Zhaowan cemetery (776 AD), in Baotou, in Inner Mongolia and the tomb of Zhang Wenzao (1093 AD) in Xiabali in northeast China (Steinhardt 1998) are among similar examples of this layout in east Asia just before the appearance of the Mongol empire.

Fig. 13

Architectural layout of pre-Mongolian tombs in east Asia and its accordance with the ground plan of the underground rock-cut monument of Maraghe (left). A: Tomb No.1 in Huatehua Cemetery. B: Plan of the Abaoji’s tomb. C: Shoroon Bumbagar tomb in Biannuur. D: tomb of Shoroon Dov Barrow. E: Tomb of Xiao Yi. F: Tangut’s royal tomb. G: Tomb of Mme Yi. H: tomb of Xu Xianxiu. I & J: Boshan’s tombs, Chifeng. K: Koguryo’s Tombs. L: Zhang Wenzao’s tomb. M: Plan of M67 at Zhaowan cemetery

The role of spaces D and E remains essentially the same in the construction of this building. If they were hewn shortly after the initial plan in the Ilkhanid era, then it is likely that the aim was to have a ground plan like the tomb of Abaoji (810 AD) (Xinlin 2010), Shoroon Bamagar (670 AD), Shoron Dov Barrow (667 AD) (Danilov et al. 2010), Xiao Xi (1018 AD) (Wen Lihe 1989: 325), or Mme Yi (960 AD) (Steinhardt 1988), all of which belong to the clans of eastern Turks in Mongolia. In these buildings, the previous arrangement is further extended by enlargement of the space by adding extra rooms on either wing of the hall that leads to the main room (room C). In addition to the mentioned example, around 33 dome tombs have been recorded form the Yuan dynasty (1271–1368) in north China all of which including an underground structure with a domical cover that reaches the surfaces of the ground and resemble the Mongol tradition in funeral architecture (Erdenebat 2009). However, despite traceable architectural ties that bind the rock-cut complex of Maraghe to these regions at that time, the different material could thus indicate the general tendency of the local architecture in using the available tuff layer. Although the overall pattern of these domes might give an impression of the Monoglian yurt of ger, there has been no evidence to support this claim. The appearance of the pigment in the main circular room in this building raises this hypothesis that it might have been a mural tomb like those at Ningjia Yingzi in Chifeng, (Xiang 1983), Houdesheng in Liangcheng Country, M1 at Fujiantun in Lingyuan Country (Dong 2018), and M2 at Sanyanjing in Chifeng, all of which are located in Inner Mongolia, in the heartland of the Mognols.

Considering architectural parallels and the appearance of two funeral cells beneath the main room, it is unfortunate that the building does not contain any inscription to trace the personalities of the buried persons. The author of the Azerbaijan’s Historic Monuments affirms that the person who was buried in this monument was a famous clergyman of the Ilkhanid era, Mullah Māsum e-Maraghei (Mokhlesi 1992: 181). However, this is not a primary source, the author’s phraseology suggests that he is speaking from hearsay. On the other hand, the title Māsum was frequently used to indicate the chastity of those who were buried in many similar functions like Imāmzādeh Māsum in Tehran and Imāmzādeh Māsum in Hamadan; therefore, we must regard the authenticity of this claim as unreliable at this point. There is, therefore, no evidence, apart from the fact that this monument was used as a burial place as the underground cells of space J and the recently plundered Zarih (Beigbabapour 2009: 179), to confirm this. It should be mentioned that the building was probably considered a holy shrine after burying or removing the body of an elite, royal family member or even a religious authority that gave a religious function to this building. However, we should keep in mind that religious buildings in the context of Islamic Iran were always subject to renomination to eradicate the non-Islamic identity of the buried person and it is likely that the same has happened with this building as well (Kiyani 2017: 108).

Once we have found traceable similarities between this building and religious functions in Mongolia, the hypothesis that the site might suggest a similar function becomes tenable. This similarity might be motivated by ethnic or social reasons, considering the deployment of the Mongol tribes from this region and their settlement in northwest Iran after 1255 AD. Although after Genghis Khan the Mongol empire divided into four states (Vernadsky 1953: 112), thanks to the lifting of borders, people, ideas, and knowledge could circulate between these khanates, helping to contribute to the transformation of the artistic and architectural heritage throughout the Mongol empire (Allsen 1996: 10). However, whether central Asian masters were present to construct this building or it was a regional architectural production requires more research.

It is also necessary to develop a comprehension of the importance of the tree in the Mongol ideology. In the Turko-Mongol mythology the tree has been considered as one of the five sacred elements of the universe and sacred trees were associated with the worship of the Mongolic god Olgen (Oraz 2017: 177). In the cosmogony of Tengrism,³⁹ in addition to the real earthly world, an upper world (heavenly realm) and an underworld, which are connected by a “navel of the world” ( world axis ). In Tengrism this navel is the so-called “World Tree” (Muler 2019: 57). In old Altaic texts, the presence of trees symbolizes creation (Inan 2016: 21) and it gave birth to the heroic characters of Deniz Khan (See), Gök Khan (Sky), and Dağ Khan (Mountain) in the folkloric legend of Oğuznama⁴⁰ (Rashīd al-Dīn 1988: 25). Other tales state that Genghis Khan had given orders to plant a tree over his tomb after his death (Rashīd al-Dīn 1988: 387). Sacred tree was the principle feature in the Mongolic sacrificial ceremonies and dancing around a tree assumed a special importance in these rituals (Oraz 2017: 177). Even after the conversion of Ghazan Khan to Islam, he adhered to these ancient traditions (Rashīd al-Dīn 1979: 41). In an anecdote form Rashīd al-Dīn in Tarikh-i Mubarak-i Ghazani, Ghazan made a pilgrimage to a lofty tree and ordered a large offering to show his thankfulness for his success and prayed a lot to the old God (Tengri) and intended to build a shrine for himself there (Rashīd al-Dīn 1974: 141).⁴¹ Considering the textual evidence of the tree as an important symbol in Mongol mythology and legend, it is logical to conclude that the tree in the central axis of this building would have been included to reflect this belief (Fig. 14).

Fig. 14

Historic old tree in the central axis of the rock-cut structure in Maraghe before 1997. (Digital collection of Cultural Heritage and Tourism Organization; Iran)

Historical sources indicate that a bias in favor of a hidden location on top of lofty mountains was given preference over normal burial customs in Mongol times, a policy hitherto known before Ghazan’s conversion to Islam (1295 CE) (Rashīd al-Dīn 1994: 997). Even if we adhere to Rashīd al-Dīn to disguise the location of the Khan, by considering any type of dome in this complex, it would be destined to be seen from afar. Furthermore, with the presence of a tree, its location could be further highlighted. Otherwise, the whole complex was arranged to attract the attention of visitors and was not likely a hidden location for a pre-Ghazan royal resting place. According to the historical text, Hulagu Khan was buried in a mountain in the island known as Shahi Island which literally translated to Royal Island (Rashīd al-Dīn 1979: 1021). This landscape has enough potential to be a sacred place due to the inaccessible mountain inside the vast lake which provided adequate reasons to select it as a royal tomb. As the location of the burial place of Arghun, Zipoli (1974) believed that he was buried in an anonymous location in the mountains of Sujas. It is widely recognized that Ghazan and Oljeito built their own magnificent domes in Ghazaniyya (Tabriz) and Sultaniyya (Zanjan), respectively. Also, Brambilla (2015) suggested that Tepe Nur is the monumental mausoleum built by the last Ilkhanid emperor, Abu Said (1305–1335 CE). The underground rock-cut complex at Maraghe is certainly not the mausoleum of any of the major viziers of Ghazan, Oljeito, or Abu Said. These viziers were all centered in Tabriz, where Rashid al-Din built a famous mausoleum and Taj al-Din Alishah a huge mosque (Mūstūfī 1983: 87; Ibn and Battūta 1953, 98). Otherwise, when it comes to intended use of this building as a royal tomb, there would be no unknown Ilkhanid elite burial candidate for this monument in this region.

8 Conclusions

The impression given by the underground rock-cut architecture at Maraghe is that it was a burial place where the Mongol funeral customs were still firmly kept. Even though we have no clear evidence for a construction date for this building, the inscriptions and decorations help to assign its construction to the 13th and 14th centuries, indicating that it enjoyed the patronage of an Ilkhanid contractor. Indeed, the peculiar combination of an underground rock-cut complex in connection with a cemetery that appeared with the advent of Ilkhanid empire in Northwest Iran, in many aspects, reminds one of the internationalization of the Mongol funeral heritage. From this point of view, the underground rock-cut monument of Maraghe indicates the multi-cultural balance within Ilkhanid society and its attribution to the Mihr Temple has been based upon wrong assumptions, and simply misconstrued interpretations.

Although the original plan of this building has been lost through the enlargement process, as mentioned earlier, both construction phases are only found to be appropriate for funeral purposes. Additionally, Maraghe was the first center of interest for the Mongol empire in Iran. If we accept the context of Ilkhanid Iran as flexible, the architectural layout of this building has much more affinity with the royal tombs in Mongolia, Northern China, and Inner Mongolia, but the stylistic variation in decoration between this monument and east Asian specimens should be used as further justification for separating them based on geographical boundaries. Otherwise, it is neither surprising or unlikely that this building may have been an attempt to try to copy the earlier Mongol construction methodologies in the Islamic context of Iran. However, it was certainly not a one-way axis of artistic transfer. Rather, the evidence suggests a thriving world of artistic exchange, and many decoration, ideas and techniques brought from Armenia. The architectural similarity between this building and some dome tombs in northern China and Mongolia invites questions about the workforce involved. Of course, this does not mean that all the manpower came to Maraghe specifically for the project; it is possible that some of the stonemasons, as well as the men who carried out heavy manual labor, were local. Although earlier scholarship has tended to focus on a pre-Islamic mithraic function, I would like to remain with the conservative idea of an Ilkhanid tomb, emphasizing the fact that diverse architectural traditions did survive in the Ilkhanid capital of Maraghe. However, owing to the insufficient archaeological excavations, some understandings are still limited, and therefore the in-depth grasp of this monument is still waiting for the supplement and examination of more archaeological discoveries.