The caves at Ellora in Maharashtra, India, are among the most impressive examples of rock-cut architecture found on the subcontinent. The Archaeological Survey of India has identified thirty-four caves at Ellora carved into an exposure of basalt that stretches over a mile in length.
The caves date from the late sixth through tenth centuries—an important period of temple building in India as regional rulers, merchants and traders, and religious communities sought to establish their presence and power through the patronage of structural and rock-cut architecture.
At Ellora, there are caves affiliated with three of India’s religious traditions: Hinduism, Buddhism, and Jainism. As in other multireligious sites that feature caves—such as Aihole and Badami in Karnataka—it is the rocky topography that likely attracted initial excavation activities. Caves were, and continue to be, associated with specific sacred mountains that are central to ancient Indian cosmologies. Thus, caves not only offered quite suitable conditions for ascetic living on a practical level, but also created a sacred environment for both occupants and visitors.
Scholarship on Ellora tends to present the site’s internal development as follows:
- early Hindu activity (c. late 6th through early 7th century C.E.)
- Buddhist phase (c. 7th through the first quarter of the 8th century C.E.)
- renewed Hindu activity (c. mid-8th through early 9th century C.E.)
- Jain phase (c. 9th through 10th century C.E.).
However, recent research reveals that the site developed more organically with artists working on Hindu, Buddhist, and Jain caves simultaneously at certain times.  Artists did not work on caves according to their own personal religious affiliation nor did they work as specialists for only one type of cave. Close examinations of architectural elements and their ornamentation—as well as treatments of sculptural form and iconographical features—indicate that artists moved back and forth across the site. Importantly, artists established sacred visual vocabularies that crossed religious boundaries.
Early caves devoted to Shiva
The earliest caves (14, 17, 19, 20, 21, 26, 27, 28 and 29) are located near the center of the site.  Almost all are dedicated to the Hindu god Shiva and house a rock-cut linga (an abstract, columnar form of Shiva).
Cave 29, the largest early cave
The largest of the early spaces is Cave 29 and it is carved next to a waterfall created by the force of Ellora’s Velganga river during the monsoon season. As seen in other expansive rock-cut sites across the subcontinent (such as the caves of Ajanta), early caves tend to be clustered close to a primary water source.
The linga shrine in Cave 29 is carved as a separate architectural unit that can be circumambulated within the cave. The presentation of a free-standing linga-shrine distinguishes Ellora’s early Shaiva caves from those that are carved in the eighth and ninth centuries (Caves 15, 16, 18, 22, 23, and 24). In the later Shaiva caves, the linga shrine is carved directly from the back wall of the cave.
In addition to the linga shrine, Cave 29 features numerous sculpted panels of Shiva and other deities in figural form. Subjects include, for example, Shiva destroying demons, Shiva as lord of yogis, the marriage of Shiva and Parvati, and Shiva dancing. Similar to the sculpted programs in Cave 1 at Elephanta, the panels appear to present Shiva through complementary acts of creation, destruction, earthly and cosmic engagements, and ascetic practices. These two colossal reliefs at Ellora are carved facing each other at the western entrance to Cave 29. The relief of Shiva destroying Andhaka presents the god with multiple arms so that he can hold a variety of weapons and other items. His dynamic lunge and ferocious facial expression emphasize the arduous task in slaying this demon. In the Shiva subduing Ravana panel, the god simply places his foot down upon the ground to quell the shaking of Mt. Kailasa caused by Ravana below. Both sculpted reliefs prominently include Shiva’s female consort Parvati.
While the subjects of these panels can be identified quite easily, interestingly, many of the carvings in Cave 29 remain “unfinished” in one way or another. One of the most incredible aspects of rock-cut architecture is the fact that artists could create colossal sculpted programs without worrying about the structural integrity of the cave. Furthermore, the subdued lighting within these spaces surely contributed to perceptions of sculpted form materializing (or self-manifesting) from temple walls. Adding to this visual experience is the fact that most of Ellora’s carved architectural and sculptural elements were once painted (or were intended to be painted) with natural pigments. Recent examinations of these reliefs by art historian Vidya Dehejia and sculptor Peter Rockwell prompt us to consider processes of sculpting as well as premodern conceptions of “completeness.”  For example, in the Shiva destroying Andhaka panel, portions of the rock beneath Parvati have been left unfinished. These areas only reveal the marks of the point chisel—a preliminary tool in sculpting. Though “incomplete”, the unfinished areas in this relief in no way detract from its readability and suggest that artists and viewers in premodern India had more fluid notions of a sculpture’s state of completion.
View of Buddhist Caves 6, 5, 5RW, and 4 at Ellora, c. 7th century (photo: VA Photography, CC BY-SA 4.0)
Buddhist caves (Caves 1–12)
In the early seventh century, artists shifted to available rock at the southern end of the site. Here, while working on Cave 14—a temple likely created for the worship of the Hindu goddess Durga—artists also began carving Buddhist residences and worship spaces. 
There are twelve Buddhist caves at Ellora: eleven serving, in part, as residences (viharas) for monastic communities and one chaitya hall containing a rock-cut stupa carved with a figural, seated Buddha. The singular chaitya hall at Ellora is apsidal in plan and is carved with a ribbed barrel-vaulted ceiling. Rock-cut pillars forming side aisles create a path for circumambulation around the stupa. In addition, devotees can approach on axis from the central aisle of the hall to engage directly with the Buddha image.
The site’s earliest viharas (Caves 2, 3, 4, 5 and 6) present a relatively standard plan encountered in the late fifth-century caves at Ajanta and in select caves at nearby Aurangabad. This includes a quadrangular, pillared hall with individual monastic cells, and a shrine containing a large rock-cut Buddha image, such as we see in Cave 3. Other sculptural programs in Ellora’s Buddhist caves include reliefs of female deities, bodhisattvas, series of Buddhas, and mandalas (sacred diagrams).
Cave 12: architecture as sacred diagram
In her important study of Ellora’s Buddhist caves, Geri Malandra identifies the mandala as a governing framework for understanding the sculptural programs and their arrangement within the caves.  Mandalas are schematic arrangements of Buddhas, Buddhist deities, and bodhisattvas. They can serve as maps or diagrams during meditation and worship to reveal complex spatial connections between earthly and cosmic realms. Ellora’s Buddhist caves feature mandalas carved in relief and as part of an expansive, three-dimensional shrine program.
Numerous reliefs of mandalas are carved in the three-storied vihara, Cave 12. In this relief, we see nine figures, each placed into its own rectangular space. The central figure is a Buddha who is surrounded by eight ornamented bodhisattvas. Mandalas like this one may have served as a guide for worship as it would have been viewed prior to entering into the main shrine. Significantly, this mandala replicates the large-scale shrine programs of all three stories of Cave 12 which include a seated Buddha flanked by groups of bodhisattvas.
Although mandalas are popularly known through painted examples across Buddhist Asia, Ellora’s Buddhist caves can also be viewed as three-dimensional representations. This is underscored by the carving of such diagrams in liminal spaces of the cave, such as in or near shrine rooms, at stairwells, and on ground level (such as the example from Cave 12). Our movement through the Buddhist caves—inward towards shrines and upward towards top levels—parallels visualization practices aided by mandalas. Some of these practices likely included visualizing the different realms governed over by specific bodhisattvas that ultimately lead to enlightenment and Buddhahood.
Questions of patronage at Ellora
Scholars generally identify the patrons of Ellora’s early Hindu and Buddhist caves as rulers of the Kalachuri and Chalukya courts, respectively, even though there is no inscriptional evidence at the site to confirm such associations. Royal use of the site is, however, recorded in three inscriptions affiliated with a dynasty of rulers known as the Rashtrakutas. In 742 C.E., the Rashtrakuta king Dantidurga visited the site and purified himself by bathing at the tirtha (sacred pilgrimage place) of the “Lord of the Cave” (a phrase from Dantidurga’s copper-plate land grants, found at Ellora, that refers to the god Shiva).  This ritual ablution may have taken place in the body of water formed by the waterfall adjacent to Cave 29. This would make sense because Cave 29 likely continued to serve as an important Shiva temple at Ellora.
The importance of Ellora for Dantidurga is also witnessed in an inscription carved on a monolithic pavilion that precedes Cave 15. Throughout Ellora’s caves, this is the only inscription carved on a monument to link it with a royal dynasty. The inscription, located above the central, pierced-stone window, begins with an invocation to Shiva and documents the genealogy of the Rashtrakutas. The list of kings ends with praise for Dantidurga’s accomplishments, including his victory over other regional kings. 
While the Cave 15 inscription ties Dantidurga (and the Rashtrakutas) to Ellora, curiously, it does not identify him as the patron of the monolithic pavilion or the multistoried cave behind it. Moreover, the visual evidence of Cave 15 indicates that this monument was initially carved out prior to Rashtrakuta involvement with the site. Images of seated Buddhas are carved on the bracket-capitals of the upper-story veranda pillars, indicating that Cave 15 began as a multistoried Buddhist cave, much like Caves 11 and 12. However, perhaps in response to Dantidurga’s visit, large Shaiva guardians were subsequently carved into the façade (and interior programs were added) to transform the cave into an abode of Shiva. This transformation ushered in a dynamic period of excavation activity at Ellora by a multitude of patrons in the eighth through tenth centuries. Monuments include additional temples to Shiva (such as the Kailasanatha) and a Jain complex of caves.
Kailasanatha—a temple beyond the cave
The site’s most impressive monument is actually not a cave per se, but a rock-cut monolithic temple. This temple is Cave 16, also known as the Kailasanatha. The name of this temple makes reference to Shiva’s residence on Mt. Kailasa—a sacred mountain and axis mundi in select Indian cosmologies. To create the temple, stone-cutters had to first carve three trenches into the hill to isolate a rectangular block of stone. While teams of artists tackled the back of the block to begin the top of the temple’s main tower, others focused on shaping the colossal gateway and other components of the temple. Remarkably, though the Kailasanatha is cut from the living rock, it replicates elements found in structural (not rock-cut) temples, including the gateway, a separate pavilion for Shiva’s bull Nandi, a pillared main hall, subsidiary shrines around the main tower, and a sanctum enshrining a rock-cut linga. The extraordinary nature of the Kailasanatha as a simulacrum of a structural temple is only surpassed by the high quality of its imagery. The temple tower and the exterior walls of the main hall are carved with a variety of proliferating forms including vegetal and aquatic motifs, vessels of abundance, flying celestial figures, and the entire pantheon of Hindu deities. A number of the colossal reliefs carved on the exterior of the temple make direct reference to Shiva’s Himalayan abode on Mt. Kailasa—further linking this monument to Shiva’s place of residence.
Carved entirely from the rock, the Kailasanatha is perhaps the most dramatic expression of the temple as sacred mountain and cave. The spectacular form and innovation of this sculpted temple is highlighted in a famous inscription issued by a ruling king of the dynasty’s branch in Gujarat. The reference to the Kailasanatha temple is found in the genealogical account of the dynasty and states that Dantidurga’s successor, Krishnaraja I constructed a temple so wondrous that it astonished the gods and even the architect who made it.  Although most scholars cite this inscription as evidence of Krishnaraja’s patronage of Ellora’s Kailasanatha temple, this particular record is somewhat problematic given its late date (forty years after his reign), place of issue (western state of Gujarat), and the fact that this information is not repeated in other Rashtrakuta genealogies. Thus, the exact nature of Rashtrakuta patronage at Ellora remains unclear. 
Through a Jain lens
Ellora’s Jain caves provide more information in regard to patronage and the site’s internal development. Five Jain caves (Caves 30–34) are carved at the northern end of the site. The sculpted and painted programs feature Jinas, Jain deities (particularly associated with health, wealth, and abundance), and the Jain figure Bahubali. Reliefs of Bahubali at Ellora typically present the figure flanked by two females who are in the process of removing the vines that encircle his legs. Overhead are flying, celestial couples who hold long flower garlands to honor and worship Bahubali who has just attained omniscience due to his steadfast meditation. The forest setting of this event is referenced through the vines and the small deer that gather around his feet.
In addition to these subjects, the Jain caves are also carved with numerous human worshippers, both monastic and lay.  Typically they are carved in gestures of homage at the feet of a Jina. Some of these figures are identified through an inscription as the patron(s) of the sculpture, such as Nagavarma and Sohila in Cave 32. Their presence points to patterns of patronage at the site that go beyond royal elites. Moreover, it appears that scenes of human devotion made an impression on later imagery carved at the site, as seen in some additions to the Kailasanatha complex that feature portraits of anonymous devotees worshipping Shiva.
While the vast majority of Ellora’s Jain caves were carved in the ninth and tenth centuries, there is some evidence that Jain artistic activity was initiated much earlier. This can be seen in the pillar type carved in the upper-story veranda of Cave 33. These pillars, with their fluted, cushion-shaped capitals and tall, plain shafts, mirror those carved in some of the site’s early Shaiva and Buddhist caves. Also supporting an earlier date for the inception of Jain activity are three Jina images carved adjacent to Cave 33. The throne bases supporting these Jinas have a cloth or fabric that drapes down and covers the central section of the base. Carved above the cloth is a double-lotus seat that visually supports the Jina. These motifs are similarly found on the thrones supporting select Buddhas at the site.  These elements, among others, indicate a seventh or early eighth-century date for initial Jain activity. This is significant as it demonstrates that teams of artists were working on Hindu, Buddhist, and Jain caves simultaneously during this period.
The three Jina images adjacent to Cave 33 likely served as “placeholders” until artists could resume work on the Jain caves after the completion of the monolithic Kailasanatha temple in the mid-eighth century. As a result, we also have the innovative “Chhota (small) Kailasa” at Ellora (Cave 30)—a monolithic Jain temple that replicates the Kailasanatha in terms of its process of creation, directional orientation, and main architectural components, albeit on a smaller scale. 
Undoubtedly, future research will reveal more evidence for the movement of artists across the site and highlight ways that Ellora functioned as a multireligious tirtha in India’s early medieval period.
 Lisa N. Owen, “Relationships between Art, Architecture, and Devotional Practices at Ellora,” in Pia Brancaccio, ed. Living Rock: Buddhist, Hindu, and Jain Cave Temples in the Western Deccan (Mumbai: Marg Publications, 2013), pp. 126-137.
 For the relative chronology of these early caves see Walter Spink, “Ellora’s Earliest Phase,” Bulletin of the American Academy of Benares I (1967), pp. 11-22.
 Vidya Dehejia and Peter Rockwell, The Unfinished: Stone Carvers at Work on the Indian Subcontinent (Delhi: Roli Books, 2016), pp. 67-73.
 Spink, “Ellora’s Earliest Phase,” pp. 17-20.
 Geri H. Malandra, Unfolding a Mandala: The Buddhist Cave Temples at Ellora (Albany: State University of New York Press, 1993).
 S.K. Dikshit, “Ellora Plates of Dantidurga: Saka 663,” Epigraphia Indica XXV (1940), pp. 25-31.
 James Burgess, Report on the Elura Cave Temples and the Brahmanical and Jaina Caves in Western India, Archaeological Survey of Western India 5, 1877-80 (London: Trubner & Co., 1883; reprint, Varanasi: Indological Book House, 1970), pp. 87-89.
 R.G. Bhandarkar, “The Rashtrakuta King Krishnaraja I and Elapura,” Indian Antiquary vol. 12 (1883), pp. 228-30.
 For a re-assessment of Rashtrakuta patronage at Ellora, see Lisa N. Owen, Carving Devotion in the Jain Caves at Ellora (Leiden: Brill, 2012), pp. 131-137.
 Owen, Carving Devotion, pp. 16-24 and 137-163.
 Owen, “Relationships between Art, Architecture, and Devotional Practices at Ellora,” pp. 130-32.
 Owen, Carving Devotion, pp. 28-36.