Mamandur
Caves

Mamandur is a village in Tiruvannamalai district of Tamil Nadu, on the Kanchipuram–Vandavasi road near Dusi. The caves are situated on the banks of the Palar River and the site is also commonly referred to as Dusi Mamandur.

The supplied study places the site roughly 10 km from Kanchipuram, 80 km from Chennai, 85 km from Vellore, 100 km from Tiruvannamalai and about 50 km from Chengalpattu.

The supplied material dates the earliest evidence of human habitation at Mamandur to the Iron Age megalithic period, based on a stone weapon discovered by scholars, suggesting settled communities engaged in resource use around approximately 1200 BCE.

A Tamil Brahmi inscription found in a cavern in the village mentions a chieftain who captured Torur and gifted the cavern, dated in the supplied material to the 2nd–3rd century CE. The hill also contains a cave shelter with Samanar padukai (Jain beds), associated in the document with Jaina monks and a chieftain named Kaniman in the 3rd–4th century CE.

According to the supplied material, Pallava king Mahendravarman I was initially a patron of Jainism before his conversion to the Shaiva tradition. Kanchipuram, his capital, is presented as a cosmopolitan centre in which Buddhist, Jain, Vaishnava and Shaiva traditions flourished in different areas.

The document describes his conversion to Shaivism in connection with Thirunavukkarasar (Appar), and places these events within the wider Bhakti movement.

Four rock-cut temples are described: the first two at Mamandur village and the other two in neighbouring Narasamangalam, also called Narasapalaiyam. The material connects the village name “Narasamangalam” with Narasimhavarman.

The first two caves are described as Mahendra-style Pallava architecture, while the next two are in Narasimhavarman style and remain unfinished, with natural rock faults identified as an important reason.

The caves were carved in hard granite and gneiss using hammer-and-chisel techniques. The document also identifies Chitramegha Tataka, the Mamandur lake behind the caves, as a tank believed to have been excavated on the orders of Mahendravarman I, significant for understanding Pallava water management and the relationship between temple complexes and hydraulic infrastructure.

Cave 1 is described as definitively attributed to Mahendravarman I because it contains his inscription. It is the northernmost of the four caves and lies north of Cave II.

The cave faces east and has an Ardha mandapa with two massive pillars and two pilasters. The pillars are divided into square saduram at the base, an octagonal kattu in the middle and square saduram above, with lotus medallions. The potika is large and curved without taranga ornaments.

The single garbhagriha is oblong and has a rock-cut plain platform on the back wall. There are no dwarapalakas. The supplied text identifies it as a Vishnu cave, and notes that its plan closely resembles the Vishnu cave at Mahendravadi. Faded green paint survives on wall surfaces, and an asymmetrical square pit in the platform may have been intended for a stucco image or painted plaque.

Cave 2 is attributed in the supplied material to Mahendravarman I, but it has no Pallava inscription. Two Chola Tamil inscriptions inside confirm that the principal deity was Shiva; one inscription calls the temple Rudravalisvaram.

The back wall has three shrine cells. The central shrine is associated with Shiva; its dvarapalas carry clubs and wear jata-bharas. The southern shrine’s dvarapalas resemble rishis holding lotuses and are presented as likely Brahma. The northern shrine’s figures stand in tribhanga with hands in adoration and patra-kundalas, and are presented as likely Vishnu.

The facade has four large pillars and four pilasters. The pillars are more massive than those of Cave I, with square sadurams at top and bottom and an octagonal kattu shaft. The corbels are angular, described as evidence of early experimental work. The kapota cornice is unfinished.

The lingam currently in the central shrine is described as a later installation; the floor socket was originally intended for a sculptural image.

The supplied material places this large cave in the second period, around the time of Narasimhavarman I (630–668 CE). It is described as the most ambitious and complex unfinished monument in the group, with five shrines without deities.

Each shrine is almost cubical, with traces of plaster. Oblong sockets near the back walls appear intended for bas-reliefs on stone slabs inserted at a later stage. There are no dvarapalakas and no inscriptions, so the intended deities cannot be identified from the supplied evidence.

The cave was intended as a pillared circumambulatory space around a line of five central shrines, essentially a colonnaded cloister surrounding the shrines. The supplied material compares the concept with the Panchapandava cave temple immediately south of Arjuna’s Penance at Mahabalipuram.

The eastern facade has seven pillars with different degrees of completion. A second row of four pillars creates outer and inner verandahs. On the back wall are five shrine cells at a higher floor level, each approached by a three-step sopana staircase with elephant-trunk-like curved side rails.

Cave 4 is the southernmost of the four caves, at the northern end of the southern hill in Narasapalaiyam village. The supplied material describes it as an early attempt in the time of Narasimhavarman I that was abandoned because of a rock fault.

The facade has two pillars and two pilasters with an oblong mandapa behind. The back wall was blocked out with chisel lines indicating three intended shrine cells.

Provision was also roughly cut for the vajana and kapota, but these remained uncut. No inscriptions have been found in this temple.

The supplied material identifies rock faults on the western end of the southern continuation of the cloister as an important reason the circuit could not be completed. The design was then improvised, with four lateral shrines added without sopanas, adhishtanas or pilasters as a rough compromise.

For Cave 4, a natural rock fault ran diagonally from the top of the southern side through the line of the front pillars. The supplied material states that this prevented the pillars from being cut to the required thickness and that the two pillars broke off at their tops along the fault.