It is among the mountainous, wooded expanses of Madhya Pradesh, on the southern peripheries of the Gupta empire, that a greater part of the earliest enduring free-standing shrines can be found. They largely belong to the late Gupta period. It is undeniably to the relative segregation of the region, south of the attack routes, that they owe their preservation. The area is rich in stone, unlike Madhyadesa, where most of the temples must have been constructed by brick, thereby destroyed effortlessly. Less than a score persists now- a little more than pile of fragments, none have preserved intact superstructures. They comprise the major part of a single cubical sanctum (garbhagriha) of well-groomed masonry, with minimal mouldings, entered by intricately carved doorways of the greatest beauty. Lying about loose round several of these shrines were embellished dados, stone grilles, waterspouts and amalakas, evidencing to a diplomatic use of sculptural ornaments in the purest Gupta style. The cult images of two of them, both Vaiava, the Narasimha, in Eran and the Vamana incarnation of Vishnu at Marhia, have been distinguished among bordering wreckages. The Saiva shrines must have housed the lingas.

The much-restored Gupta temples at Sand, the Kankali Ma at Tigowa, together with the partially rock-cut Cave no.1 at Udayagiri (Vidisa), are possibly the earliest of the lasting shrines. Their porticoes with four columns in antis, the distance between the middle columns wider than the others, plain doorways, and in the case of the first two, what may have been flat roofs, puts them in a class apart from the others. But it would be wrong to hypothesise any historical growth from these sparse survivals to the later Gupta shrines. Sand is plausibly the earliest (first quarter of the 5th century) due to its bell capitals; the others possess the vase and foliation character.

The persisting temples belong to the latter half of 5th century and 6th. Dates are completely in lack, and a chronology must be established on stylistic grounds only. The temples possessed an upper shrine or a superstructure (shikhara) of some kind. At this stage in some cases, plausibly abbreviated- like the Marhia temple, without a porch, seems to have had only two low bhiimis (storeys or tiers). Evaluating by the antarapatra (here a band of panels in the recesses between the bhiimis of bhiimiprasdada- late Kakinga-type temples) and gavasaks elements, which have survived, Bhumara may have had a shikhara of the same kind. Most of the temples were raised upon a jagati - a podium. The more sophisticated Parvati temple at Nachna Kuthara, changed almost beyond identification by misleading renovation, had an internal circumfusing passage and an upper shrine room. Sculpture of variable quality was established at the site. The Dasavatara at Deogarh, whose superstructure is largely hypothetical and the Bhitargaon temple near Kanpur, the lonesome survivor of the multitudinous brick shrines, which must have been erected in Madhyadesa during Gupta times, had higher, curvilinear shikharas. Also outside Madhya Pradesh, the largely ruined shrine at Mukundarra, south of Kotah-, which presumably had an attached front hall-, boasted internal columns and pilasters, a feature unique in Gupta structural temples. It seems to have united brick walls with carved stone pillars and capitals and other exceedingly classy ornamental elements.

If there is any conclusion to be drawn from this meager, haphazard collection of shrines preserved by the mere chance of endurance (to which must be added the much reinstated Mahabodhi temple at Bodh Gaya) it can be found in the conflicts between them, which would seem to suggest that during the Gupta period the temple was still in a decisive stage. Many more types, which have left no successors, must have been built, a hypothesis strengthened by the inimitable features of the recently revealed Devarani temple at Tali. In the same process, sarvatobhadra shrines (with entrances on all four sides), so admired in later periods, must certainly have been built during Gupta times, and yet none have lasted.

In quality of sculptural ornamentation nevertheless and predominantly in the unexcelled exquisiteness of their doorways, these little shrines personify the Gupta aesthetic at its prime. As a mark of its inventiveness and originality, it is indeed an appealing idea to unsophisticate the entire external walls of the Nachna Kuthara Parvati temple, with animals in the fissures and crevices. Such a conduct is securely based on two major aspects of Shiva, as the Lord of Kailash amongst the rocky crests of the Himalayas and as the sovereign ascetic in wilderness. At Bhitargaon and Deogarh, the images have already begun their move to the walls of the temple. In Deogarh, three walls are panelled with vistas from Vaishnava myth- Narayana, Anantasayana (Vishnu asleep on the serpent Ananta) and the saving of Indra`s elephant (Gajendramokra). Magnificient as are these last two themes- and the figure of the slumbering Vishnu, dreaming another anon (kalpa) of creation into existence, is one of the most controlling of all Indian iconographic conceptions- the firmness and power of early Gupta sculpture is conceding to a softer, more fragile and finally weaker in style. Idiosyncrasies include the well-known stylised postures of the men below the sleeping Vishnu and even a note of disrespect can be distinguished in the haughtiness of the god as he arrives to the liberate the elephant from the labours of the lotus stems.

A petite number of sculptures belonging to the late Gupta period have been found in Mandasor, in nearby Sondni and in Kilchipura, in a territory of Madhya Pradesh in Rajasthan, but a division of Malwa in earlier times. The colossus Shiva stele, its face re-cut in the earlier part of 20th century, is environed by massed galas, some of them grotesques, resonant of the hosts of Mara in some Gandhara stele. The architectural settings of an exquisite air of dvarapalas at Sondni, the site of Yasodharman`s pompous inscription of 533-4, have distinctive post-Gupta characteristics. The flattening of the relief in the Kilchipura stambha also harbingers the impending style. On the other hand, the glowing pucker-lines at the top of some of the dhotis, and sometimes even the small turn-over, are trademarks of Gupta fashion.

A much superior body of sculpture from northern Gujarat and southern Rajasthan, in a prodigious assortment of styles, once more beats chronology by its total lack of dates. From the main sash of some of the male figures from Samalaji, a secondary loop hangs down, as sometimes at Parel and Elephanta. The iconography is well formulated. An abnormal number of these primarily small figures are in the round. Dates in the 6th early 7th centuries seem most likely. Among mother and child groups (skamdamatas- the god Skanda or Kartikeya had six origins or mothers) the finest were from Tanesara-Mahadeva (Rajasthan). Their rhetorical antecedents are complex to pin down. Gently moulded and in rather mawkish postures, they are possibly more easily assimilable to western taste than any other Indian sculptures, partially because of their summoning of the Virgin and Child.

If it is hard to envision the original background of much of the stone sculpture, it is even more difficult in the case of the terracotta`s- some of substantial size and many of them relief plaques, yet they rank among the artistic glorifications of the Gupta period. In an associated medium, the five stucco sculptures of the inimitable Maniyar Math at Rajgir, presumably a Naga Temple, were placed in niches, like so much later works. Only the brick temple at Bhitargaon yields a glance of the process in which terracotta sculpture became a fundamental part of the temple. The large relief panels in the niches of the main walls have all but fallen apart, but the upper storeys are still equipped (among other subjects) with grotesques (cf. the gargoyles of Gothic cathedrals) and makara friezes. Of an elegance and sophistication infrequently excelled elsewhere in the world are the two large river goddesses, 4ft. 8in. (1.47m.) in height, from the excavated remnants of a huge brick temple at Ahichchhatra. Many tiny terracotta statuettes, much worn and placeable as Gupta, predominantly by their hairstyles, have been discovered in Rajghat (Varannasi) and additional sites.

Scientific excavation of the stupa at Devnimori in Gujarat, ahead of a reservoir project, divulged, along with rows of perceptively modelled terracotta Buddhas, a repertoire of architectural ornamentation, replicating and often excelling that in stone, including maybe the earliest version of the brilliant laterally cut foliation friezes of the Gupta period, in addition to acanthus friezes and elaborate gavaksas. Another stupa further west, near Mirpur Khas in Sind and a century or more later in date, has also generated large terracotta reliefs of Buddhas and a great deal of decorative detail, in a drier style compared to Devnimori. The Mirpur Khas stupa was the first to have a buried chamber on one side of its base.