Pramāṇa · Search · About · API

The Age of the Imperial Guptas c8: Chapter IV. Architecture

The Age of the Imperial Guptas c8

Language
English

Chapter IV. Architecture

c8.2

Specimens of civil and domestic architecture of the period are still very rare, in spite of the excavations carried on by the Indian Archæological Department at Rājagṛiha, Pāṭaliputra, Vaiśālī, Benares, Bhiṭā, Kosām, and Kurukṣetra. Marshall’s dating of the buildings at Bhiṭā is very appealing but totally unscientific. The assignment of such early dates to a building simply because a single seal or stone-axe was found in a chamber is unconvincing. The amount of material discovered by the late Drs. Bloch and Spooner at Vaiśālī prove that the chamber in which the seals were discovered belong to the Gupta period along with the connected buildings of the same series and stratum, but nothing is left of such buildings today and even the bricks of the walls excavated have been carried away. Similarly, the Gupta structures discovered by Spooner in the uppermost stratum of his excavations at Pāṭaliputra have also disappeared. Even when they were just unearthed they conveyed very little to the student of the history of our architecture.

c8.3

There remains to be discussed the architecture of the sacred buildings of the period of which a few examples have survived up to our times.

c8.4

Even at the present day, there is considerable divergence of opinion amongst scholars regarding the true type of temples of Gupta period proper. Earlier writers were of opinion that like most modern and mediaeval temples, temples of the Gupta type also possessed *Śikharas* or spires. The most prominent examples cited by Cunningham and other earlier writers are the great brick temples at Bhitargaon in the Cawnpore district and Mahābodhi or Bodh-gaya in the Gaya district. The same class of writers described other Gupta temples, *e.g.*, those at Sanchi, Tigowa in the Jubbulpur district and at Bodh-Gaya as being flat-roofed. Recent discoveries of Gupta temples have proved that the Gupta temple proper did not possess a *Śikhara*. In the case of the Mahābodhi temple at Bodh-Gaya the arguments produced for assigning it to the Gupta period are: that it was seen by the Chinese pilgrim Yuan Chwang, that the dimensions¹ given by him agree with those of the present temple before its repairs, that the Ceylonese king Meghavarṇa sent an embassy to Samudragupta asking for a permission to build a monastery at Bodh-Gaya², and finally the Chinese pilgrim I-Tsing states that the Mahābodhi *Vihāra* was built by the Ceylonese. Chavannes has proved from Chinese records that the Mahābodhi *Vihāra* was built by a king of Ceylon. “Near the Bodhi tree was the Mahābodhi Vihāra built by a king of Ceylon.”¹ Writers like Mr. E. B. Havell think that the Mahābodhi temple was built in the first century B.C.² After a thorough examination of the entire structure of the Mahābodhi temple the writer could find no trace, both inside and outside, of decorative motifs of the Gupta period in any part of it. On the other hand all other temples whether Buddhist or Hindu, which can be definitely assigned to the Gupta period on the ground of epigraphy, always show the use of decorative motifs distinctive of the Gupta period, e.g., the original main Shrine and the Gupta monasteries at Sarnath, the Hindu temple at Muṇḍeśvarī near Bhabua in the Arrah district, the Gupta temple at Budh-Gaya, the temple of Pārvatī at Nachna-Kuthara in the Ajaygadh State, the temple of Śiva at Bhumara in the Nagod State, the later Gupta temple at Deogadh in the Jhansi district and the Gupta temple at Sanchi in the Bhopal State.

c8.5

¹ Coomaraswamy—History of Indian and Indonesian Art, p. 81.

c8.6

² Smith—Early History of India, 4th Edition, p. 304.

c8.7

¹ Takakusu—I-Tsing, quoting Chavannes, Memoirs, p. 84; p. xxxii and Note 2.

c8.8

² A Study of Indo-Aryan Civilisation, p. 100.

c8.9

The original outline of the Mahābodhi temple was of a different shape, which was changed when it was encased in fresh masonry at the time of its repairs (1880–92). This slim outline of the Śikhara can be seen in earlier photographs.³ The outline of the original Śikhara along with the absence of particular decorative *motifs* of the Gupta period proper prove that the Mahābodhi temple could not have been erected earlier than the 8th century A.D. There are two other temples of the same type and probably of the same date in South Bihar, one of which, though not repaired, is still in better preservation than the original Mahābodhi temple. These are the temple of Śiva at Konch, near Tikari, in the Gaya district and the great Buddhist *Vihāra* at Nālandā in the Patna district. The latter was excavated and partly destroyed by the late Mr. A. M. Broadley, I.C.S., when most of the carved stones were removed to the Museum founded by him at Bihar whence they were removed to the Calcutta Museum in 1897-98. The ruins have been re-excavated recently by the Archæological Survey of India. In this temple the *Śikhara* collapsed long ago. According to an inscription discovered on the door-jamb of this temple it was rebuilt in the 11th year of the reign of Mahīpāla I of Bengal *i.e.*, towards the close of the 10th century A.D. The *motifs* employed here and at Konch prove that the present Mahābodhi temple can not be earlier than the 8th century A.D.

c8.10

³ R. L. Mitra—Buddha-Gaya, pl. XV; Cunningham—Mahabodhi, pl. XXXI

c8.11

The great brick temple at Bhitargaon in the Cawnpore district lies twenty miles to the south of Cawnpore town. The temple was first described by Cunningham in 1875-76 and he assigned it to the Gupta period.¹ It was more accurately surveyed in recent times by Dr. J. Ph. Vogel, formerly a Superintendent of the Archæological Survey of India. Dr. Vogel states, “The outer ornamentation of terra cotta sculpture is certainly the most striking feature of the Bhitargaon temple (Plates IV and V). The walls rise in bold mouldings, their upper portions being decorated with a row of rectangular panels alternating with ornamental pilasters. It has been noticed above that the early plinth of the Nirvāṇa temple at Kasia is embellished in a very similar fashion and that on that account there is good reason to ascribe the Bhitargaon temple to the early Gupta period.”¹ The exterior walls of this temple were decorated with terra cotta panels in niches, proving that it did not belong to the type of the sancta at Bhumara or Nachna Kuthara. Moreover, there is a *Śikhara* in which there is a series of arches which are not exactly *Caitya*-windows.² The plinth mouldings are totally absent proving the difference between it and the temples at Nachna Kuthara and Bhumara.³ Therefore, its affinity lies more with the temple at Deogadh than with those of Nachna Kuthara and Bhumara because of the presence of the stumpy *Śikhara*.⁴ The art of the terra cotta panels of the Bhitargaon temple have been assigned to the Gupta period without sufficient reasons. A closer comparison with the products of the three great schools of Gupta Art in India e.g., Mathurā, Benares, and Pāṭaliputra prove that the terra cotta panels of Bhitargaon are later in date than the great Deogadh panels or the finer bas-reliefs of Sarnath. The mediaeval art of the United Provinces still remains to be studied. Sufficient remains of the periods of Bhoja I, Adivāraha and his son Mahendrapāla I, have been discovered all over the United Provinces but these scattered remains have not been analysed yet. Inscribed records of Bhoja I¹ on sculpture exist at Deogadh and Pehoa and some of Mahīpāla I at Asni in the Fatehpur district. At the same time the *Śikhara* of the Bhitargaon temple and its decorative motifs bear a curious resemblance to those of the temple of the Somavaṁśī kings in the Raipur district of the Central Provinces.²

c8.12

¹ Archæological Survey Reports Vol. XI, pp. 40-46.

c8.13

¹ *Annual Report of the Archæological Survey of India, 1908-9, p. 9.*

c8.14

² *Ibid., pl. I.*

c8.15

³ *Memoirs of the Archæological Survey of India, No. 16. pl. II.*

c8.16

⁴ *Cunningham—Archæological Survey Report, Vol. XI. pl. XV.*

c8.17

¹ Ibid., Vol. X, pl. XXXIII.

c8.18

² Ibid., Vol. XVII, pl. XVI; E. B. Havell—A Study of Indo-Aryan Civilization pl. XXXV, L, and LI.

c8.19

The temples of the early, later and post-Gupta periods are enumerated below:—

c8.20

I. Early Gupta (319–550 A.D.):—

c8.21

1. The temple of Śiva at Bhumra, about six miles from Unchehra railway station on Jubbulpur-Itarsi Section of the G.I.P. Ry., discovered by the writer in 1920.

c8.22

2. The earlier temple of Śiva at Nachna Kuthara, called temple of Pārvatī by Cunningham, in the Ajaygadh state, about 10 miles from Bhumra, described by Cunningham and by the writer in 1919.

c8.23

3. The temple called “Lad Khan’s temple” at Aihole in the Bijapur district of the Bombay Presidency, built in the Early Cālukya times.

c8.24

II. Later Gupta (551-605 A.D.):—

c8.25

4. The later Gupta temple at Deogadh in Jhansi district, generally mistaken to be an Early Gupta structure.

c8.26

III. Post-Gupta (606-700 A.D.):—

c8.27

5. The small Post-Gupta temple at Śaṅkargadh in the Nagod State, discovered by the writer in 1920.

c8.28

6. The Post-Gupta temple at Nachna Kuthara in the Ajaygadh district, discovered by Cunningham and described by the writer in 1919.

c8.29

7. The Post-Gupta temple at Mundeśvarī, near Bhabua in the Arrah or Shahabad district of Bihar and Orissa.

c8.30

Though no large inscription has been discovered in the temples at Deogadh, Bhumra, and Nachna Kuthara, the dates of these three temples can be accurately deduced from short inscriptions and mason’s marks. From these data we can safely deduce that the flat-roofed temples of Bhumra and Nachna Kuthara belong to the Early Gupta period, which ended in the middle of the 6th century A.D. These two temples must be described before considering the Later and Post-Gupta temples.

c8.31

In the case of both of these temples the architect was concerned more with the provision of a covered path of circumambulation (pradakṣiṇā-patha) as at Elephanta, than with a Śikhara. In fact, though there is a small chamber above the main shrine in the case of the Early Gupta temple at Nachna Kuthara, there is no indication of any Śikhara in the case of both of these structures. In both cases the flat roof of the sanctum or the chamber above it indicates that there was no Śikhara. The earlier temple at Nachna Kuthara possesses a small flat-roofed chamber above the sanctum, proving thereby that the architect did not intend to build a Śikhara over this shrine. These two temples prove that the origin of the Śikhara or spire in Indian temple architecture is much later than the period of domination of the Early Gupta Emperors in Northern India. From its style as well as the mason's marks in the temple of Śiva discovered at Bhumra, this one is the earlier of the two. The remains of this shrine were excavated by the writer in 1920-21. The entire shrine was 35' square. In front of this square area was the plinth of the Maṇḍapa or porch measuring 29'-10" by 13'. There is a flight of steps in front of this porch, on each side of which were discovered the plinths of two smaller shrines, measuring 8'—2" by 5'—8". In the centre of the square portion of the plinth is the sanctum or *Garbha-griha*, measuring 15'—6" square, built of finely dressed Kaimur red sandstone without any mortar and roofed with long flat slabs. The rest of the space in the square area, which enclosed the sanctum, was a covered path of circumambulation as can be proved from the analogy of the exactly similar existing structures at Nachna Kuthara and Aihole, which are lighted by stone-windows of pierced screens on the sides.

c8.32

The earlier temple at Nachna Kuthara is practically of the same size as that at Bhumra. The sanctum in this case measures 15'—6" on the outside and 8' inside. The larger chamber, or path of circumambulation, which encloses the sanctum is 33' on the outside and 26' in the interior. The *Mandapa* in this case measures 26' by 12'. The steps at Nachna Kuthara are 18' by 10' while those at Bhumra are 11'—3" by 8'—5". The masonry in the case of both temples is finely coursed ashlar. The difference between these two temples lies in the detailed and exceedingly fine decorations of the Bhumra temple compared with which that at Nachna Kuthara is much plainer. While the path of circumambulation and the porch of the Bhumra temple are completely ruined, the entire structure at Nachna Kuthara was in a much better state of preservation in 1919. There is another point of difference between these two temples, which is the absence of any structure over the sanctum in the case of the temple at Bhumra. The only example of a flat roofed shrine surrounded by a covered path of circumambulation in which there is no structure over the sanctum, the temple at Bhumra, is unique in this respect. Coomaraswamy places the temple of Lad Khan at Aihole, without sufficient reason in *circa* 450 A.D. Beyond the resemblance with the earlier temple at Nachna Kuthara in having a small square cell above the sanctum and a covered path of circumambulation around the latter, lighted by large windows of pierced screens, there is no other reason to place the date of the erection of this temple earlier than the time of Kīrttivarman I of Badami, *i.e.*, the first half of the 6th century A.D.

c8.33

The Early Gupta Temple was therefore, in type, a flat roofed sanctum, with a covered path for circumambulation, having an open porch in front decorated with pure Gupta motifs. It is not possible for us to determine how this type came to be copied in the 6th century A.D., at Badami; but the designs survived in the Malabar country, up to the end of the 16th century. On the Malabar coast, especially in the modern districts of North and South Kanara, square shrines, surrounded by one or more covered paths of circumambulation, have been discovered in very large numbers, from Mudabidri near Mangalore to Gersoppa and Bhatkal in the north. In the case of all of these temples, there is no *Śikhara* over the sanctum, but the excessive rainfall of the districts demanded that the slab-roof should be sloping instead of being flat. Hence the roofs of the sancta as well as the single or double paths of circumambulation are made of stone slabs, placed in a slanting position like tiled roofs of modern buildings. This particular type of temples resembles the Early Gupta Type in many particulars; *e.g.*, the want of a *Śikhara*, one or more covered paths of circumambulation, a small open porch in the centre of the *façade* and want of ornamentations on the exterior. These temples in the north and south Kanara district are Hindu and Jaina. The Jain temples are called *Bastis* and some of them contain big and elaborate establishments. The general decline of Jainism along the Malabar coast has caused the desertion of many of these *Bastis*, but due to the munificence of the Vijayanagara emperors and the chiefs of Sugandhā or Sonda, the majority of Hindu temples are in a much better state of preservation, comparatively. The example of the great Jaina *Basti* at Bhatkal in the extreme south of the North Kanara district of the Bombay Presidency may be cited. In this case the roof of the sanctum, the path of circumambulation and the porch are all sloping and constructed of long thick plain slabs of stone finely dove-tailed and placed on heavy stone beams. The exterior is severely plain but the interior shows an amazing mass of decorative carving of the south Indian type which is quite different from the decadent Hoyśala motifs of Hampe, Penukonda, Chandragiri and Udayagiri. Another feature of these Malabar temples is that in the majority of cases they are built on stone piles, having empty spaces under the floors. There is a lamp post (*dīpa-stambha*) in each of them, which is a monolithic pillar surmounted by a stone lantern, standing apart from the building. The sloping roofs of the Jaina *Basti* of Bhatkal decrease gradually in height; the roof of the sanctum being the highest, next to it comes that of the first path of circumambulation, then comes the second path of circumambulation, while that of the porch is the lowest. It is impossible at the present day to find out how the Early Gupta-Temple-type came to survive in a modified form in the extreme south-west of the Indian peninsula and survived there for eleven centuries after the Gupta period. Some links have been discovered recently in the country between East Central India and the Malabar Coast. Though these links possess *Śikharas*, their plan consists of a small *Garbha-gṛha* enclosed by a covered path of circumambulation. Such are the temples of Samgameśvara, Mallikārjuna ¹ and Virupāksha ² at Pattadkal in the Bijapur district of the Bombay Presidency. A similar arrangement exists in the Meguti temple¹, the Huchchimalli-gudi², temple No. 9³ at Aihole and in the two temples at Mahākūṭa⁴ in the same district. The second link is to be found in the group of Chandella temples at Khajuraho in the Chhattarpur State near Nowgong, slightly S.S.W. of Bhumra and Nachna Kuthara. In the Khajuraho temples, in one or two cases, a small passage has been left between the *Garbha-gṛha* and the three *Ardha-maṇḍapas* on three sides of it.

c8.34

¹ Cousens—*Chalukyan Architecture of the Kanarese districts*, pl. XXXV.

c8.35

² *Ibid*, pl. XLV.

c8.36

¹ *Ibid.*, pl. IV.

c8.37

² *Ibid.*, pl. XIII.

c8.38

³ *Ibid.*, pl. XVI.

c8.39

⁴ *Ibid.*, pl. XXVII.

c8.40

Out of the two early Gupta temples, the one at Nachna Kuthara was in a better state of preservation and from its analogy we can guess that the path of circumambulation at Bhumra also was devoid of detailed ornamentation in the interior. The exterior in this case also may have been ornamented and the slabs discovered during the excavations, consisting of sunkel panels divided by pilasters, with figures of *gaṇas* or goblins alternately, may have formed the exterior decoration.⁵ The very great number of ornamental sculptures discovered during the excavation of the temple at Bhumra proves that the *Maṇḍapa* at that place was much more elaborately decorated than the existing porch at Nachna Kuthara. This *Maṇḍapa* possessed at least one finely decorated gateway.

c8.41

⁵ *Memoirs of the Archæological Survey of India*, No. 16; the *Temple of Śiva at Bhumra*, pp. 8-10; pl. IX-XI.

c8.42

Four fragments of this gateway were recovered and on them we see the shaft of a pilaster, shaped like the rough bark of the date-palm. A parallel band bears on it that exquisitely fine arabesque which is characteristic of pure Gupta work. Miniature dwarfs are turning somersaults at the corners of the bases of the pilasters.¹ The roof of the interior of the porch was supported by graceful tapering pillars, embedded in foliated vases, the shafts of some of which are plain and some fluted.² Set against the plain ashlar masonry of the inner wall of the path of circum-ambulation and the sanctum were numerous pilasters with plain or octagonal shafts but ornamented with square bosses bearing some of the finest arabesque medallions, ever discovered in India, as well as *Kīrtimukhas*.³ The zenith of artistic excellence is reached in the case of the ornate slabs of the flat roof, many of which were recovered in a wonderful state of preservation. One of them bears on its surface a mass of arabesque foliage with a long stem, clinging to it are superb little *Amorini*; on the other we see a giant creeper with huge corrugated leaves reminiscent of Acanthus leaves of Corinthian architecture, and on a third the representations of huge waves with breaking crests.⁴ These roofing slabs are of different sizes and on others, which are narrower, we find arabesque which is relatively mediocre compared with the three described above. There are others in which leaves are arranged in squares as well as geometrical patterns.¹ Very probably the porch of Maṇḍapa was open on three sides and the lower part of its sides was composed of decorated slabs. The exterior of the porch was decorated with a line of indescribably fine Caitya-windows, containing round medallions with figures of Hindu gods and goddesses. These Caitya-windows can be crossly divided into four classes; (a) according to size and (b) according to ornamentation. Larger and smaller Caitya-windows were probably placed alternately along the cornice. The larger Caitya-windows are ornamented along the circumference of the medallions either with (i) arabesque or (ii) two small lotuses.² The medallions contain figures of Gaṇeśa, Brahmā, Yama, Kubera, Kārtikeya, Śiva dancing, Śiva seated on his bull, Sūrya, Devī as Mahisha-mardinī and Kāma. In one or two cases the smaller Caitya-windows contain either small figures of dancing Amorini or full lotus rosettes. From analogy it appears that such Caitya-windows were placed alternately according to size. Such is the position of these decorative figures on the cornice of the so-called Dharmarāja’s ratha at Mamallapuram in the Chingleput district of the Madras Presidency.³ Exactly similar but smaller *Caitya*-windows have been discovered in the early Gupta temple at Nachna Kuthara.¹

c8.43

¹ *Ibid.*, pl. IV.

c8.44

² *Ibid.*, pl. VI.

c8.45

³ *Ibid.*, pl. V.

c8.46

⁴ *Ibid.*, pl. VII.

c8.47

¹ Ibid., pl. VIII.

c8.48

² Ibid., pls. XII—XIV.

c8.49

³ Havell—A Study of Indo-Aryan civilization. The ancient and mediaeval architecture of India. London, 1915, p. 87, fig. 36.

c8.50

¹ Cunningham—*Archæological Survey Reports*, Vol. XXI, pl. XXVI.

c8.51

The masonry of both of the early Gupta temples is finely coursed small ashlar. The architect did not provide for the extremes of expansion and consequently all stones of the surface are either badly cracked or chipped at the corners, a characteristic to be found in the *Daśāvatāra* temple at Deogadh in the Jhansi District, the unfinished temple on the mound at Nemawar, on the Narmadā, in the Indore State² and in the *Śaiva* monastery at Chandrehe in the Rewa State.³ No mortar has been used in the construction of any of these two temples, nor was clamping resorted to. There was no great weight upon the walls and pillars of these two temples as the height was not much and the roofs all flat. The collapse of these two temples is due entirely to the cracking of the lintels and displacement of the foundations by tropical vegetation.

c8.52

² *Progress Report of the Archæological Survey of India, Western Circle, for the year ending 31st March, 1921*, pp. 102-6, pl. XXVII.

c8.53

³ *Ibid.*, pp. 83-5, pl. XIV.

c8.54

Towards the close of the 6th century A.D., a new member was added to the top of the flat roof of the sanctum. This was the beginning of the *Śikhara* in Northern India. The earliest example of this new member in Indian Architecture is to be found in the later Gupta temple at Deogadh in the Jhansi district. The photograph published by Cunningham in 1875 shows the remains of this new member, decorated with *Caitya*-windows and other distinctly Gupta decorative motifs.¹ Other temples of the same period are those discovered by the present writer at Sankargadh,² in the Nagod State and by Cunningham at Nachna Kuthara,³ in the Ajaygadh State. The temple at Sankargadh is earlier than the later temple at Nachna Kuthara. The date of this small temple can be fixed by a comparison of its carved door-frame with that in the later Gupta temple at Deogadh. During a recent visit to Deogadh the writer found that the plinth of the Daśāvatāra temple has been fully excavated. The plan of the temple is slightly different from that given by Cunningham. It is now certain that in this temple also there was a covered path of circumambulation, one beam of which is still sticking out; but there were four entrances to it instead of one, and all of them were provided with small porches and staircases. There were four small temples at the four corners which were probably capped by small *Āmalakas*, many of which were discovered during the excavations. On each side of each of the staircases there was a niche, only one of which is nearly entire.

c8.55

¹ Cunningham—*Archæological Survey Reports*, Vol. X, pl. XXXV.

c8.56

² *Progress Report of the Archæological Survey of India, Western Circle, for the year ending 31st March, 1920*, pp. 104-5, pl. XXVII.

c8.57

³ *Ibid.*, 1919, p. 61, pl. XVII.

c8.58

Surrounding the main temple there were a number of smaller shrines, the plinth levels of which are much lower than that of the Daśāvatāra temple. But they appear to be of a later date from some carving on some of them. One of them bears on it a row of square rosettes, placed alternately between inverted stepped pyramids, a design too common in the mediæval Jain temples on the top of the hill at Deogadh. The excavation of the area surrounding the Daśāvatāra temple has revealed many interesting facts unknown to Cunningham and earlier writers. The first is that of large pillars with the regular Gupta decorative motifs of half and three-quarters medallions on the shaft and foliated vases at the top or the bottom. Several such have been discovered, it appears at different times, but the one which lies by the side of the new sculpture-shed nearest to the Daśāvatāra temple bears on it a longish inscription in two lines in characters of the late sixth century. Evidently such pillars were used in the four porches and the temple cannot be much earlier than 575 A.D. Another interesting feature discovered during the recent excavations is the presence of a number of vignettes in the medallions of *Caitya*-windows. One of them is very clearly a vignette within another, while the second shows a door-way or niche within a vignette. The number of *Caitya*-windows discovered prove the truth of the author's previous theories about the employment of these members in the flat-roofed temples of Bhumra and Nachna Kuthara. In addition to the figures inside the medallions discovered at these two places, the new Deogadh-finds include the well-known Sarnath motif, of twin columns with cruciform capitals inside the medallions of Caitya-windows.¹

c8.59

¹ D. R. Sahni, Catalogue of the Museum of Archæology at Sarnath, pp. 254-58.

c8.60

The importance of the Deogadh temple lies in its Śikhara. The Śikhara of the Daśāvatāra temple is low with gradual curves in it as in the case of the temples of Paraśurāmeśvara and Mukteśvara at Bhuvanesvara, the twin temples at Gandharāḍi in the Baudh State and a number of Early Cālukyan temples at Aihole and Pattadkal in the Bijapur district of the Bombay Presidency. Only the lower portion of the Śikhara of the Daśāvatāra temple remains, but the entire contour can be judged and determined from it. Surrounding the base of this low pyramidal Śikhara was the flat roof of the path of circumambulation, the edge of which was decorated with large and small Caitya-windows with medallions as are to be found on the edge of the free-standing rathas at Mamallapuram in the Chingleput district. Two large pillars stood in the centre of each facade, supporting the porch and on its sides rose a short wall, on which were placed dwarf pillars and pilasters supporting the flat roof. Against each of these two walls on the sides of each porch rose in relief a slender but fairly tall niche containing some divine figure. Above the line of the topmost plinth-moulding were used large thick slabs of stone bearing stumpy pilasters alternated with figures of *gaṇas*. The smaller temples in the four corners most probably contained the *bassi relievi* discovered in previous years.¹

c8.61

¹ Coomaraswamy, *History of Art in India and Indonesia*, pl. XLIV, fig. 167.

c8.62

In the remaining portions of the *Śikhara*, the decorative *motif* predominating, is the *Caitya*-window. They were, however, much more distinct in Cunningham’s time than at the present day. This is the beginning of the long history of this *motif* which became unrecognisable in its later stylized form.

c8.63

In its front facade the entrance to the sanctum is fitted with a door-frame of the same style as those at Bhumra and Nachna Kuthara. We find the same divine figures, larger in number, at the bottoms of the door-jambs, which consist of more than one upright; the continuation of some of the arabesques and superimposed panels on the lintel, the false extension of the sides of the lintel to give an idea of massivity and finally the addition of a large boss in the centre of the lower part of the lintel bearing a fine bas-relief. In the *Śiva* temple at Bhumra this boss bears in relief a fine bust of *Śiva*, the place of which is taken in the Daśāvatāra temple by a small figure of Nārāyaṇa.

c8.64

seated on the coils of Ananta or Śeṣa. The representation of Viṣṇu of this particular style are extremely rare and no other specimen is known to us in Northern India. The only other specimen known to the writer is to be found in the great Vaiṣṇava cave-temple (No. IV) at Badami in the Bijapur district of the Bombay Presidency.¹

c8.65

¹ *Memoirs of the Archaeological Survey of India*, No. 25, *pl.* XVII *a*.

c8.66

The Daśāvatāra temple presents another new feature in its arrangement and decorative details of the three great niches containing the great bas-reliefs. Here we see for the first time the familiar Gupta *motifs* in actual use. Each niche is contained in a sunken panel formed by two uprights and a horizontal beam but most of the decorations, in all cases, go above this beam. In the case of the upright, the decorations running parallel to them, are continued in the interior but not outside. These uprights are pilasters of the familiar Gupta type : square at the bottom, plain in the lower half, bearing a narrow band of arabesque on an elevation at the bottom and the centre. In the upper half ornaments are more profuse and beginning with the three-quarters medallion containing a bas-relief it ends with a dwarf-foliated vase bearing a square abacus. Above this is a cruciform capital, partly Indo-Persepolitan in style, as below each arm one can see lions couchant in the approved and well-known method. The medallions of the pilasters contain in the case of the niche containing the great Śeṣa-śāyin, a four armed Gaṇeśa seated on the left and a two-armed figure on the right. Inside the area enclosed by these two pilasters there are two more uprights, a pair against each pilaster, both of which bear arabesque in the interspaces of meandering creepers, which is continued in the two lower bars of the lintel. A third bar is interposed below the real lintel, supported on cruciform capitals, and in this case also, the ornament is slightly variegated arabesque. On the great lintel itself, we see one of those extremely nimble artistic ideas executed in stone, consisting of a creeper with foliage, entirely unnatural, but which, somehow or other, looks very natural, with playful Amorini in the centre. Above the real lintel rises another tier with alternated sunken panels. It looks, from a distance, as if short uprights have been placed on the horizontal beam to support the sloping roof of a temple. The sunken panels bear niches or doorways carved in very shallow relief and the surface ones real lion’s heads or Kīrttimukhas. The sloping roof above is of the same shape as the Maṇḍapas (Jagamohanas) of early mediæval Orissan temples like the Paraśurāmeśvara¹ at Bhuvaneśvara or the twin temples at Gandharāḍi² in the Baudh State. A row of small dentils appear below the roof but over it is continued one of the principal architraves of the temple. It is brought over the roof of each of the niches by two advancing recessed corners. The ornamentations consist of a row of sunken panels, oblong in shape, containing niches in miniature and separated from each other by a number of flat pilasters bearing either, three long pilasters, or a mass of arabesque resembling small Acanthus leaves.

c8.67

¹ Monmohan Ganguly, *Orissa and her remains ancient and mediæval*. pl. XII.

c8.68

² *Journal of the Bihar & Orissa Research Society*, Vol XV, pp. 73-80 pl. I-II.

c8.69

The ornamentation of the niches on the three sides of the Daśāvatāra temple differs and it is apparent that the temple must have been thoroughly repaired sometime in the 7th or 8th century A.D. It is in these later additions that a definite connection can be traced between the early Gupta decorative style of niches and the slightly later Gurjjara-Pratihāra style, a very good example of which was found by Pundit Govind Mālaviya to the proper right of the great Jaina temple on the low hill to the west of the Daśāvatāra temple.¹ The original idea underlying the addition of this new member over the flat roof of the sanctum must have been to accentuate or emphasize the position of the sanctum and to distinguish it from the rest of the structure. The corners of the Śikharas of the temples at Deogadh and Sankargadh show a slight curvature, which is to be found in the earliest mediæval temples of Orissa; in the Paraśurāmesvara temple at Bhuvaneśvara and the twin temples at Gandharāḍi in the Baudh State. In these the height of the Śikhara from the point of its junction with the side walls is exactly one-and-a-half of the side walls. Unaccountably the same proportion is to be observed in the brick temples at Sirpur, in the Central Provinces which do not belong to the 6th century as Coomaraswamy¹ supposes but to the 8th century according to the Sirpur inscription of the Somavaṁśī kings.² The Śikhara became a regular feature of Indian temples from the beginning of the 7th century A.D., though even so late as the 10th, an upper chamber continued to be erected in certain cases; such as the one on the top of the sanctum of the Buddhist temple (No. 45) at Sanchi.³

c8.70

¹ Cunningham—Archæological Survey Reports, Vol. X, pl. XXIV.

c8.71

¹ A History of Art in Indica, and Indonesia, pl. LI, Fig. 186.

c8.72

² Epigraphia Indica, Vol. XI, pp. 184-97.

c8.73

³ Annual Report of the Archæological Survey of India, 1913-14, Part II, pl. XXII.

c8.74

There are a number of temples, also belonging to the Gupta period, which are generally taken to represent the type of temples of this period, because they were the earliest to be discovered. Such are the small temples of Bodh-Gaya, in the Gaya district, Tigowa, in the Jubbulpur district, and Sanchi, in the Bhopal State. These temples consist of a square sanctum with a small porch or verandah in front of it. The temple at Tigowa is probably earlier in date than the Gupta period proper, because the pillars and pilasters of the verandah bear Indo-Persepolitan capitals. The size of the small temples at Sanchi and Bodh-Gaya prove that they were auxiliary temples and not principal shrines. But because they were the earliest known temples of the Gupta period, they came to be called the representative type of temples of the Gupta period. Recent discoveries have proved that the Gupta temple at Bodh-Gaya, to the right of the passage as one gets out of the doorway of the Great Temple, is perhaps the same as that built by the Ceylonese. The Gupta temples at Sanchi and Bodh-Gaya are Buddhist temples or shrines, the type of which had just lost its originality, because at this stage both Buddhist and Jain temples were beginning to be unified with the Hindu or Brahmanical temple type.

c8.75

The oldest known temples of Northern India, belonging to the earlier part of the 7th century A.D., are the second temple at Nachna Kuthara and the temple of Mundeśvari, near Bhabua, in the Shahabad or Arrah district of Bihar and Orissa. At Nachna Kuthara the second temple possesses a fine Śikhara. It lies to the south-west of the early Gupta Temple and enshrines one of the largest four-faced Lingas (Caturmukha-Mahādeva) ever discovered. In front of the sanctum there was a small porch of twelve pillars which had collapsed when the present writer saw it in 1919. The sanctum is a plain square cell without an Antarāla, on the top of which is a modest Śikhara. The upper part of the Śikhara, consisting of the pinnacle, is damaged. From a distance this temple looks like other temples of the mediæval period of Northern India (800–1200 A.D.), but a closer scrutiny reveals certain features, which tend to prove that this temple must be very closely allied in date to the fifth and sixth century temples described in the preceding pages. The most important feature is the door-frame. The resemblance of the plastic work and the arrangement of its door-frame to that at Bhumra is so complete that it is extremely difficult to believe that they are even slightly different in date. The arabesque of the first band with the climbing Amorini, the flying figures of the lintel and the false recesses at the ends of the lintel containing female figures all point out to a contiguous building period and not a distant one. There was no path of circumambulation in this case and the exterior of the walls of the sanctum are perfectly finished and decorated with small niches. The stone door-frame of the sanctum is one of the very rare examples of 7th century art discovered up-to-date in Northern India. The presence of the river goddesses, Gaṅgā and Yamunā, in the case of the early Gupta temples at Bhumra, Besnagar,¹ Deogadh² and the earlier temple at the same place show that it was a constant feature in Gupta architecture proper. But these figures are absent at the bottom of the jambs of the post-Gupta temple at Nachna Kuthara, though they are present in the Mundesvari temple. The stone door-frame of the 8th century temples at Dhamtari and Sirpur in the Central Provinces are slightly different from that of the post-Gupta temple of Nachna Kuthara.

c8.76

¹ Coomaraswamy—History of Art in India and Indonesia, pl. XLVII, fig. 177.

c8.77

² Cunningham—Archaeological Survey Reports, Vol. X, pl. XXVI.

c8.78

The temple of Mundesvari, discovered by the late Dr. T. Bloch in 1905-6, has not drawn that amount of attention, which it deserves. It was actually in existence in the Harṣa year 30 = 635-6 A.D. The only person who has mentioned it in the history of architecture is Coomarswamy; but even he has omitted to publish its photograph at the time of its discovery. Bloch discovered fragments of a votive inscription one section of which was discovered by the late Mr. P. C. Mukharji as early as 1891. This inscription was inscribed on a separate pillar and records a donation by a noble named Bhāgudalana in the reign of a minor prince named Udayasena, who held the titles of *Mahāsāmanta*, *Mahāpratīhāra*, and *Mahārāja* and who was evidently a subordinate chief under Harṣa. This inscription mentions the erection of a temple (*Maṭha*) close to the temple of Vinīteśvara and certain donations to a temple of Viṣṇu called Maṇḍaleśvara. The present name Muṇḍeśvarī is evidently a corruption of the ancient Maṇḍaleśvara.¹ This temple underwent considerable changes during the Pāla period (800–1200 A.D.) when many additions and alterations were carried out. Though the *Sikhara* has disappeared and in the place of the large smooth ashlar work the Public Works Department of the British Government has crowned it with a low parapet of undressed or roughly dressed masonry, sufficient indications remain to prove that this building is a Post-Gupta structure. The most noticeable feature in the Mundeśvarī temple is its plan; it is neither square nor circular, but hexagonal. Temples with such plans are extremely rare in India. The decorations consists of a broad round moulding at the base of the plinth of the same type as that at Bhumra and the stone-work of the earliest main shrine at Sarnath, near Benares. The plinth is further decorated with *Kīrttimukhas*, with garlands hanging as loops from their mouths, and tassels at the junctions of the loops. The carvings of the great stone door-frame consists of bands of arabesque work, superimposed panels containing single or double human figures, and other well recognised Gupta decorative motifs. The carvings of the pillars and pilasters of the windows, bearing neat *Chaitya-windows* on them, all proclaim the Mundeśvarī temple to be a direct descendant of Gupta temples proper, yet not very far removed from the age of the Early Guptas. As at Nachna Kuthara and Bhumra, as well as Aihole and Pattadkal, pierced stone-frames admit light into the interior and even the bottoms of the jambs of such window-frames bear the figures of the river goddesses, Gaṅgā and Yamunā. Fragments of bas-reliefs discovered by Mr. J. C. French, I.C.S., exhibit the general decadence of Post-Gupta plastic art.

c8.79

¹ *Epigraphia Indica*, Vol. IX, pp. 282-3.

c8.80

Gupta architects evolved a temple-type in which there was provision for a covered path of circumambulation surrounding the sanctum, which was closed on three sides and which provided one entrance only and an open porch in front. The subsequent modifications of this temple type are beyond the limit of our discussion but in the Post-Gupta period we find the addition of a modest Śikhara without any other important changes. Excavations at Mundeśvarī and Śankargadh may yet reveal traces of a covered path of circumambulation.

Linked texts (35)