After all, however, the subject is one more suited to the purposes of the mythologist and the sculptor than to the architect. Like all rock-cut examples, except the Dravidian, the caves have the intolerable defect of having no exteriors, and consequently no external architectural form. The only parts of them which strictly belong to architectural art are their pillars, and though a series of them would be interesting, they vary so much, from the nature of the material in which they are carved, and from local circumstances, that they do not possess the same historical significance that external forms would afford. Such a pillar, for instance, as this one from the cave called Lanka, on the side of the pit in which the Kylas stands (Woodcut No. 242), though in exquisite taste as a rock-cut example, where the utmost strength is apparently required to support the mass of rock above, does not afford any points of comparison with structural examples of the same age. In a building it would be cumbersome and absurd; under a mass of rock it is elegant and appropriate. The pillars in the caves at Mahavellipore fail from the opposite fault: they retain their structural form, though used in the rock, and look frail and weak in consequence; but while this diversity in practice prevailed, it prevents their use as a chronometric scale being appreciated, as it would be if the practice had been uniform. As, however, No. 3 at Badami is a cave with a positive date, A.D. 579, it may be well to give a plan and section (Woodcuts Nos. 243 and 244) to illustrate its peculiarities, so as to enable a comparison to be made between it and other examples. Its details will be found fully illustrated in Mr. Burgess’s report.

Though not one of the largest, it is still a fine cave, its verandah measuring 70 ft., with a depth of 50 ft., beyond which is a simple plain cell, containing the lingam. At one end of the verandah is the Narasingha Avatar; at the other end Vishnu seated on the five-headed serpent Ananta. The front pillars have each three brackets, of very wooden design, all of which are ornamented by two or three figures, generally a male and female, with a child or dwarf—all of considerable beauty and delicacy of execution. The inner pillars are varied, and more architectural in their forms, but in the best style of Hindu art.

Compared with the style of art found at Amravati, on the opposite coast, it is curious to observe how nearly Buddha, seated on the many-headed Naga,[456] resembles Vishnu on Ananta in the last woodcut, and though the religion is changed, the art has hardly altered to such an extent as might be expected, considering that two centuries had probably elapsed between the execution of these two bas-reliefs. The change of religion, however, is complete, for though Buddha does appear at Badami, it is in the very subordinate position of the ninth Avatar of Vishnu.[457]

Sometimes the Hindus successfully conquered one of the main difficulties of cave architecture by excavating them on the spur of a hill, as at the Dhumnar Lena at Ellora, or by surrounding them by courts, as at Elephanta; so that light was introduced on three sides instead of only one, as was too often the case both with Buddhist and Hindu excavations. These two, though probably among the last, are certainly the finest Hindu excavations existing, if looked at from an architectural point of view. The Ellora example is the larger and finer, measuring 150 ft. each way (Woodcut No. 245). That at Elephanta, though extremely similar in general arrangement, is less regular in plan, and also somewhat smaller, measuring only 130 ft. by 120 ft. It is easy to see that if these temples stood in the open they would only be porches, like that at Baillûr (Woodcut No. 221), and numberless other examples, which are found everywhere; but the necessities of rock-cut architecture required that the cella should be placed inside the mantapa, or porch, instead of externally to it, as was always the case in structural examples. This, perhaps, was hardly to be regretted; but it shows how little the practice of cutting temples in the rock was suited to the temple-forms of the Hindus, and we need not, therefore, feel surprised how readily they abandoned it when any idea of rivalling the Buddhists had ceased to prompt their efforts in this direction.

So far as I know, there is only one example where the Indo-Aryan architects attempted to rival the Dravidian in producing a monolithic exterior. It is at a place called Dhumnar, in Rajputana, where, as already mentioned (ante, p. 162), there is an extensive series of late Buddhist excavations. In order to mark their triumph over that fallen faith, the Hindus, apparently in the 8th century, drove an open cutting into the side of the hill, till they came to a part high enough for their purpose. Here they enlarged this cutting into a pit 105 ft. by 70 ft., leaving a temple of very elegant architecture standing in the centre, with seven small cells surrounding it, precisely as was done in the case of the Kylas at Ellora. The effect, however, can hardly be said to be pleasing (Woodcut No. 246). A temple standing in a pit is always an anomaly, but in this instance it is valuable as an unaltered example of the style, and as showing how small shrines—which have too often disappeared—were originally grouped round the greater shrines. The value of this characteristic we shall be better able to appreciate when we come to describe the temples at Brambanam and other places in Java. When the Jains adopted the architecture of the Buddhists, they filled their residential cells with images, and made them into little temples, and the Hindus seem to some extent to have adopted the same practice as here exemplified, but never carried it to the same extent.

With a sufficient number of examples, it would be easy to trace the rise and fall of this cellular system, and few things would be more interesting; for now that we find it in full force in the Buddhist monasteries at Gandhara (ante, p. 171), it would be most important to be able to say exactly when the monk made way to the image. In India Proper there is no instance of this being done in Buddhist times, or before A.D. 650, and hitherto we have been in the habit of considering it a purely Jaina arrangement. This must now be modified, but the question still remains—to what extent should this be done?

One more illustration must conclude what we have at present to say of Hindu rock-cut temples. It is found near Poonah, and is very little known, though much more appropriate to cave architecture than most examples of its class. The temple itself is a simple pillared hall, with apparently ten pillars in front, and probably had originally a structural sikra built on the upper plateau to mark the position of the sanctuary. The most original part of it, however, is the Nundi pavilion, which stands in the courtyard in front of the temple (Woodcut No. 247). It is circular in plan, and its roof—which is a great slab of rock—is supported by, apparently, sixteen square pillars of very simple form. Altogether it is as appropriate a bit of design as is to be found in Hindu cave architecture. It has, however, the defect only too common in those Hindu excavations—that, being in a pit, it can be looked down upon; which is a test very few buildings can stand, and to which none ought to be exposed.

CHAPTER IV.

CENTRAL AND NORTHERN INDIA.

None of these temples—none, at least, that are now complete—seem to be of any great antiquity. At Erun, in the Saugor territory, are some fragments of columns, and several sculptures that seem to belong to the flourishing age of the Guptas, say about A.D. 450; and in the Mokundra Pass there are the remains of a choultrie that may be as old, or older, but it is a mere fragment,[458] and has no inscription upon it.

Among the more complete examples, the oldest I know of, and consequently the most beautiful, is the porch or temple at Chandravati, near Jahra Puttun, in Rajputana. In its neighbourhood Colonel Tod found an inscription, dated A.D. 691,[459] which at one time I thought might have been taken from this temple, and consequently might give its date, which would fairly agree with the style,[460] judged from that of some of the caves at Ellora, which it very much resembles. As recent discoveries, however, have forced us to carry their dates further back by at least a century, it is probable that this too must go back to about the year 600, or thereabouts. Indeed, with the Chaöri in the Mokundra Pass, and the pillars at Erun, this Chandravati fragment completes the list of all we at present can feel sure of having been erected before the dark ages. There may be others, and, if so, it would be well they were examined, for this is certainly one of the most elegant specimens of architecture in India (Woodcut No. 248). It has not the poetry of arrangement of the Jaina octagonal domes, but it approaches very nearly to them by the large square space in the centre, which was covered by the most elegantly designed and most exquisitely carved roof known to exist anywhere. Its arrangement is evidently borrowed from that of Buddhist viharas, and it differs from them in style because their interiors were always plastered and painted; here, on the contrary, everything is honestly carved in stone.[461]

Leaving these fragments, one of the oldest, and certainly one of the most perfect, in Central India is the now desecrated temple at Barrolli, situated in a wild and romantic spot, not far from the falls of the Chumbul, whose distant roar in the still night is the only sound that breaks the silence of the solitude around them. The principal temple, represented in the Woodcut No. 249, may probably be added to the list of buildings enumerated above as erected before 750 A.D. It certainly is at least a century more modern than that at Chandravati, and, pending a more precise determination, may be ascribed to the 8th or 9th century, and is one of the few of that age now known which were originally dedicated to Siva. Its general outline is identical with that of the contemporary Orissan

temples. But instead of the astylar enclosed porch, or mantapa, it has a pillared portico of great elegance, whose roof reaches half-way up the temple, and is sculptured with a richness and complexity of design almost unrivalled, even in those days of patient prodigality of labour. It will be observed in the plan (Woodcut No. 250) that the dimensions are remarkably small, and the temple is barely 60 ft. high, so that its merit consists entirely in its shape and proportions, and in the elegance and profusion of the ornament that covers it.

In front of the temple is a detached porch, here called a Chaöri, or nuptial hall (the same word, I believe, as Choultrie in the south), in which tradition records the marriage of a Huna (Hun) prince to a Rajputni bride, for which purpose it is said to have been erected;[462] but whether this is so or not, it is one of the finest examples of such detached halls known in the north. We miss here the octagonal dome of the Jains, which would have given elegance and relief to its ceiling, though the variety in the spacing of the columns has been attained by a different process. The dome was seldom if ever employed in Hindu architecture, but they seem to have attempted to gain sufficient relief to their otherwise monotonous arrangement of columns by breaking up the external outline of the plan of the mantapa, and by ranging the aisles diagonally across the building, instead of placing them parallel to the sides.

The other two temples here are somewhat taller and more pointed in their form, and are consequently either more modern in date, or if of the same age—which may possibly be the case—would bring the date of the whole group down to the 10th century, which, after all, may be their true date, though I am at present inclined to think the more ancient date more consistent with our present knowledge.

A little way from the great temple are two pillars, one of which is here represented (Woodcut No. 251). They evidently supported one of those torans, or triumphal archways, which succeeded the gateways of the Buddhist topes, and form frequently a very pleasing adjunct to Hindu temples. They are, however, frail edifices at best, and easily overthrown, wherever the bigotry of the Moslems came into play.

Gualior.

One temple, existing in the fortress of Gualior, has been already described under the title of the Jaina Temple (ante, p. 244), though whether it is Jaina or Vaishnava is by no means easily determined. At the same place there is another, bearing the not very dignified name of the Teli ka Mandir, or Oilman’s Temple (Woodcut No. 252). It is a square of 60 ft. each way, with a portico on the east projecting about 11 ft. Unlike the other temples we have been describing, it does not terminate upwards in a pyramid, nor is it crowned by an amalaka, but in a ridge of about 30 ft. in extent, which may originally have had three amalakas upon it. I cannot help believing that this form of temple was once more common than we now find it. There are several examples of it at Mahavellipore (Woodcuts Nos. 181, 182), evidently copied from a form common among the Buddhists, and one very beautiful example is found at Bhuvaneswar,[463] there called Kapila Devi, and dedicated to Siva. The Teli ka Mandir was originally dedicated to Vishnu, but afterwards converted to the worship of Siva. There is no inscription or any tradition from which its date can be gathered, but on the whole I am inclined to place it in the 10th or 11th century.

Khajurâho.

As mentioned above, the finest and most extensive group of temples belonging to the northern or Indo-Aryan style of architecture is that gathered round the great temple at Bhuvaneswar. They are also the most interesting historically, inasmuch as their dates extend through five or six centuries, and they alone consequently enable us to bridge over the dark ages of Indian art. From its remote situation, Orissa seems to have escaped, to a great extent at least, from the troubles that agitated northern and western India during the 8th and 9th centuries; and though from this cause we can find nothing in Central India to fill up the gap between Chandravati and Gualior, in Orissa the series is complete, and, if properly examined and described, would afford a consecutive history of the style from say 500 to 1100 or 1200 A.D.

Next in interest and extent to the Bhuvaneswar group is that at Khajurâho,[464] in Bundelcund, as before mentioned (p. 245). At

this place there are now to be found some thirty important temples, all of which, with the exception of the Chaonsat Jogini and the Ganthai, described when treating of Jaina architecture, are of the same or nearly the same age. Nor is it difficult, from their style and from the inscriptions gathered by General Cunningham, to see what that age was. The inscriptions range from A.D. 954 to A.D. 1001; and though it is not clear to what particular temple they apply, we shall not probably err much if we assign the whole twenty-eight temples he enumerates to the century beginning 950 and ending 1050, with a margin of a few years either way. What renders this group more than usually interesting is, that the Khajurâho temples are nearly equally divided between the three great Indian religions: one-third being Jaina, one-third Vaishnava, and the remainder Saiva; and all being contemporary, it conveys an impression of toleration we were hardly prepared for after the struggles of the preceding centuries, though it might have been expected three centuries earlier.

A curious result of this toleration or community of feeling is, that the architecture of all the three groups is so similar that, looking to it alone, no one could say to which of the three religions any particular temple belonged. It is only when their sculptures are examined that their original destination becomes apparent, and even then there are anomalies which it is difficult to explain. A portion, for instance, of the sculptures of the principal Saiva temple—the Kandarya Mahadeo—are of a grossly indecent character;[465] the only instance, so far as I know, of anything of the sort being found in a Saiva temple, that bad pre-eminence being reserved to temples belonging to the worshippers of Vishnu. It is possible that it may originally have belonged to the latter sect; but, taking all the circumstances into consideration, this is most unlikely, and the fact must be added to many others to prove how mixed together the various sects were even at that time, and how little antagonistic they then were to each other.

The general character of these temples may be gathered from the annexed representation (Woodcut No. 253) of the principal Saiva temple, the Kandarya Mahadeo. As will be seen from the plan (Woodcut No. 254), it is 109 ft. in length, by 60 ft. in breadth over all, and externally is 116 ft. above the ground, and 88 ft. above its own floor. Its basement, or perpendicular part, is, like all the great temples here, surrounded by three rows of sculptured figures. General Cunningham counted 872 statues on and in this temple, ranging from 2½ ft. to 3 ft. in height, or about half life-size, and they are mixed up with a profusion of vegetable forms and conventional details which defy description. The vimana, or tower, it will be observed, is built

up of smaller repetitions of itself, which became at this age one of the favourite modes of decoration, and afterwards an essential feature of the style. Here it is managed with singular grace, giving great variety and play of light and shade, without unnecessarily breaking up the outline. The roof of the porch, as seen in front, is a little confused, but as seen on the flank it rises pleasingly step by step till it abuts against the tower, every part of the internal arrangement being appropriately distinguished on the exterior.

If we compare the design of the Jaina temple (Woodcut No. 136) with that of this building, we cannot but admit that the former is by far the most elegant, but on the other hand the richness and vigour of the Mahadeo temple redeem its want of elegance and fascinates in spite of its somewhat confused outline. The Jaina temple is the legitimate outcrop of the class of temples that originated in the Great Temple at Bhuvaneswar, while the Kandarya Mahadeo exhibits a complete development of that style of decoration which resulted in continued repetition of itself on a smaller scale to make up a complete whole. Both systems have their advantages, but on the whole the simpler seems to be preferable to the more complicated mode of design.

Udaipur.

The examples already given will perhaps have sufficed to render the general form of the Indo-Aryan temple familiar to the reader, but as no two are quite like one another, their variety is infinite. There is one form, however, which became very fashionable about the 11th century, and is so characteristic that it deserves to be illustrated. Fortunately a very perfect example exists at a place called Udaipur, near Bhilsa, in the Bhopal territory.

As will be seen from the Woodcut (No. 255) the porch is covered with a low pyramidal roof, placed diagonally on the substructure, and rising in steps, each of which is ornamented with vases or urns of varying shapes. The tower is ornamented by four flat bands, of great beauty and elegance of design, between each of which are thirty-five little repetitions of itself, placed one above the other in five tiers, the whole surmounted by an amalaka, and an urn of very elegant design. As every part of this is carved with great precision and delicacy, and as the whole is quite perfect at the present day, there are few temples of its class which give a better idea of the style than this one. Fortunately, too, its date is perfectly well known. From an inscription copied by Lieutenant Burt, it appears it was erected by a king who was reigning at Malwa, in the year 1060 of our era.[466]

At Kallian, in Bombay harbour, there is a temple called Ambernath, very similar to this, on making drawings and casts from which the Bombay government has lately spent a good deal of money.[467] It is, however, in a very ruinous state, and even when perfect could never have been equal to this one at Udaipur, and to many others on which the money might have been better laid out. In it there is a slab with an inscription, dated in the Saka year 782, or A.D. 860.[468] It is not quite clear, however, whether this inscription belongs to the temple which we now see, or to an earlier one, fragments of which are found built into the vimana of the present one. If the date of the temple is that just quoted, as Dr. Bhau Daji would have us believe, all that can be said is that it is utterly anomalous. If it is in A.D. 1070, as another inscription he quotes found near the place might lead us to infer,[469] it accords with all else we know of the style.

One other illustration must complete what we now have to say regarding these Indo-Aryan temples. It is one of the most modern of the style, having been erected by Meera Baie, the wife of Khumbo Rana of Chittore (A.D. 1418-1468). Khumbo was, as is well known, devoted to the Jaina faith, having erected the temple at Sadri (Woodcut No. 133), and the Pillar of Victory (Woodcut No. 143); yet here we find him and his wife erecting in their capital two temples dedicated to Vishnu. The king’s temple, which is close by, is very much smaller than this one, for which his wife gets credit. In plan, the only peculiarity is that the pradakshina, or procession-path round the cell, is here an open colonnade, with four little pavilions at the four corners, and this is repeated in the portico in the manner shown in the annexed diagram (Woodcut No. 256).

The roof of the portico, in the form of a pyramid, is placed diagonally as at Udaipur, while the tower itself is of so solid and unbroken an outline, that it might at first sight be ascribed to a much earlier date than the 15th century (Woodcut No. 257). When, however, it is closely looked at, we miss the frequent amalaka bands and other ornamental features of earlier times, and the crowning members are more unlike those of ancient temples. The curve, too, of its outline is regular from base to summit, and consequently feebler than that of the older examples; but taking it all in all, it certainly is more like an ancient temple than any other of its age I am acquainted with. It was a revival, the last expiring effort of a style that was dying out, in that form at least.

Vishveshwar, Benares.

If you ask a Brahman of Benares to point out to you the most ancient temple of his city, he inevitably leads you to the Vishveshwar, as not only the most holy, but the oldest of its sacred edifices. Yet it is known, and cannot be disputed, that the temple, as it now stands, was erected from the foundation in the last century, to replace one that had been thrown down and desecrated by the bigot Aurungzebe. This he did in order that he might erect on the most venerated spot of the Hindus his mosque, whose tall minarets still rear their heads in insult over all the Hindu buildings of the city. The strange thing is, that in this assertion the Brahmans are not so very

far from representing the true state of the case. There is hardly any great city in Hindustan that can show so few evidences of antiquity as Benares. The Buddhist remains at Sarnath hardly can be said to belong to the city, and even there they are, as above explained, the most modern examples of their class in India. The fact is, that the oldest buildings in the city are the Moslem tombs and buildings about the Bukariya Kund, and they almost certainly belong to the 15th century. Even the temple of Vishveshwar, which Aurungzebe destroyed, was not erected before the reign of his predecessor Akbar. The style is so nearly identical with that of known buildings of his reign, at Muttra and elsewhere, that there can be no doubt on this head. When desecrated it was the principal, and probably the most splendid, edifice of its class in the city. It may be, and probably is true, that the Vedic Brahmans erected their fire altars, and worshipped the sun, and paid adoration to the elements on this spot 4000 years ago. It may be also that the emblem of Siva has attracted admiring crowds to this spot for the last 1000 years; but there is no material evidence that before the time of Akbar (A.D. 1556-1605) any important permanent building was ever erected there to dignify the locality.

The present temple is a double one: two towers or spires almost exactly duplicates of each other. One of these is represented in the preceding woodcut (No. 258), and they are connected by a porch, crowned by a dome borrowed from the Mahomedan style, which, though graceful and pleasing in design, hardly harmonises with the architecture of the rest of the temple. The spires are each 51 ft. in height, and covered with ornament to an extent quite sufficient even in this style. The details too are all elegant, and sharply and cleanly cut, and without any evidence of vulgarity or bad taste; but they are feeble as compared with the more ancient examples, and the forms of the pyramidal parts have lost that expression of power and of constructive propriety which were so evident in the earlier stages of the art. It is, however, curiously characteristic of the style and place, that a building, barely 50 ft. in length, and the same in height, should be the principal temple in the most sacred city of the Hindus, and equally so that one hardly 150 years old should be considered as the most ancient, while it is only that which marks this most holy spot in the religious cosmogony of the Hindus.

Temple of Scindiah’s Mother, Gualior.

One more example must suffice to explain the ultimate form which the ancient towers of the Orissan temples have reached in the present century. It is just finished, having been erected by the mother of the present reigning Maharajah of Gualior, and to it has been added a tomb or cenotaph either by herself or her son. As will be seen from the woodcut (No. 259) it is elegant, though feeble as compared with ancient examples. The Mahomedan dome appears in the background, and the curved Bengali roof in the pavilion in front. The most striking peculiarity of the style is, that the sikras have nearly lost the graceful curved form, which is the most marked peculiarity of all the ancient examples. As has already been remarked, the straight-lined pyramid first appears in the Takht-i-Suleiman’s temple in Kashmir, where its introduction was probably hastened by the wooden straight-lined roofs of the original native style. It is equally evident, however, in a temple which Cheyt Sing, the Raja of Benares, erected at Ramnugger in the end of the last or beginning of the present century. Since that time the tendency has been more and more in that direction, and if not checked, the probability is that the curve will be entirely lost before the century is out. To an European eye, accustomed only to our straight-lined spires, that may seem hardly a matter for regret; but to any one educated in Eastern forms it can scarcely appear doubtful that these spires will lose half their charm if deprived of the graceful curved outline they have so long retained.

Bindrabun.

In order not to interrupt the story of the gradual development of the style, the history has been brought down to the present day in as nearly a consecutive manner as possible, thus anticipating the dates of several temples. It seems expedient, however, in any history that this should be done, for few things of its class are more interesting than to trace the progressive changes by which the robust form of the Parasurameswara temple at Bhuvaneswar, or of the great temple there, became changed into the feeble elegance of the Vishveshwar or Gualior temples. The few examples that can be adduced in such a work as this may not suffice to make this so clear to others as it is to myself. With twenty or thirty examples it could be made self-evident, and that may one day be done, and this curious chapter in architectural history be thus added to the established sequences which every true style of art affords. Meanwhile, however, it is necessary to go back a little to mention one or two aberrant types which still are not without interest.

As mentioned above, it does not appear proven that the Moslems did wantonly throw down the temples of the Hindus, except when they wanted the materials for the erection of mosques or other buildings. But, whether this was so or not, it is evident that the first three centuries of Mahomedan rule in India were singularly unfavourable for the development of Hindu art in any part of the country where their rule was firmly established. With the tolerant reign of Akbar, however, a new state of affairs was inaugurated. Not only was he himself entirely devoid of religious bigotry, but most—or at least the most eminent—of his ministers and friends were Hindus, and he lent an attentive ear to the Christian missionaries who frequented his court. But, besides its tolerance, his reign was marked by a degree of prosperity and magnificence till then unknown during that of any other Indian sovereign of his faith. Not only are his own buildings unrivalled in their extent and magnificence, but he encouraged all those around him to follow his example, and found, among others, a most apt imitator in the celebrated Man Singh of Ambêr, afterwards of Jeypore, who reigned A.D. 1592-1615. He erected at Bindrabun a temple, which either he left unfinished at his death, or the sikra of which may have been thrown down by Aurungzebe. It is one of the most interesting and elegant temples in India, and the only one, perhaps, from which an European architect might borrow a few hints.

The temple, as it now stands, consists of a cruciform porch, internally nearly quite perfect, though externally it is not clear how it was intended to be finished (Woodcuts Nos. 260, 261). The cell, too, is perfect internally—used for worship—but the sikra is gone; possibly it may never have been completed. Though not large, its dimensions are respectable, the porch measuring 117 ft. east and west, by 105 ft. north and south, and is covered by a true vault, built with radiating arches—the only instance, except one, known to exist in a Hindu temple in the north of India. Over the four arms of the cross the vault is plain, and only 20 ft. span, but in the centre it expands to 35 ft., and is quite equal in design to the best Gothic vaulting known. It is the external design of this temple, however, which is most remarkable. The angles are accentuated with singular force and decision, and the openings, which are more than sufficient for that climate, are picturesquely arranged and pleasingly divided. It is, however, the combination of vertical with horizontal lines, covering the whole surface, that forms the great merit of the design. This is, indeed, not peculiar to this temple; but at Bhuvaneswar, Hullabîd, and elsewhere, the whole surface is so overloaded with ornament as to verge on bad taste. Here the accentuation is equal, but the surfaces are comparatively plain, and the effect dependent on the elegance of the profile of the mouldings rather than on the extent of the ornamentation. Without elaborate drawings, it would be difficult to convey a correct impression of this; but the annexed view (Woodcut No. 262) of a balcony, with its accompaniments, will suffice to illustrate what is meant. The figures might as well be omitted: being carved where Moslem influences had long been strong, they are the weakest part of the design.

The other vaulted temple, just alluded to, is at Goverdhun, not far off, and built under the same tolerant influence during the reign of Akbar. It is a plain edifice 135 ft. long by 35 ft. in width externally, and both in plan and design singularly like those early Romance churches that are constantly met with in the south of France, belonging to the 11th and 12th centuries. If, indeed, the details are not too closely looked into, it might almost pass muster for an example of Christian art at that age,[470] while except in scale the plan of the porch at Bindrabun bears a most striking resemblance to that of St. Front at Perigeux (Woodcut No. 328, vol. i.). The similarity is accidental, of course; but it is curious that architects so distant in time and place should hit so nearly on the same devices to obtain certain desired effects.

Kantonuggur.

In addition to the great Indo-Aryan style of temple-building described above, there are a number of small aberrant types which it might be expedient to describe in a more extensive work; but, except one, none of them seem of sufficient importance to require illustration in a work like the present. The exceptional style is that which grew up in Bengal proper on the relaxation of the Mahomedan severity of religious intolerance, and is practised generally in the province at the present day. It may have existed earlier, but no examples are known, and it is consequently impossible to feel sure about this. Its leading characteristic is the bent cornice, copied from the bambu huts of the natives. To understand this, it may be as well to explain that the roofs of all the huts in Bengal are formed of two rectangular frames of bambus, perfectly flat and rectangular when formed, but when lifted from the ground and fitted to the substructure they are bent so that the elasticity of the bambu, resisting the flexure, keeps all the fastenings in a state of tension, which makes a singularly firm roof out of very frail materials. It is the only instance I know of elasticity being employed in building, but is so singularly successful in attaining the desired end, and is so common, that we can hardly wonder when the Bengalis turned their attention to more permanent modes of building they should have copied this one. It is nearly certain that it was employed for the same purposes before the Mahomedan sovereignty, as it is found in all the mosques at Gaur and Malda; but we do not know of its use in Hindu temples till afterwards, though now it is extremely common all over northern India.

One of the best examples of a temple in this style is that at Kantonuggur, twelve miles from the station at Dinajepore. It was commenced in A.D. 1704 and finished in 1722.[471] As will be seen from the annexed illustration (Woodcut No. 263), it is a nine-towered temple, of considerable dimensions, and of a pleasingly picturesque design. The centre pavilion is square, and, but for its pointed form, shows clearly enough its descent from the Orissan prototypes; the other eight are octagonal, and must, I fancy, be derived from Mahomedan originals. The pointed arches that prevail throughout are certainly borrowed from that style, but the building being in brick their employment was inevitable.

No stone is used in the building, and the whole surface is covered with designs in terra-cotta, partly conventional, and these are frequently repeated, as they may be without offence to taste; but the bulk of them are figure-subjects, which do not ever seem to be repeated, and form a perfect repository of the manners, customs, and costumes of the people of Bengal at the beginning of the last century. In execution they display an immeasurable inferiority to the carvings on the old temples in Orissa or the Mysore, but for general effect of richness and prodigality of labour this temple may fairly be allowed to compete with some of the earlier examples.