As remarked above, although we cannot assert with absolute certainty that the Buddhists never employed a true arch, this at least is certain--that no structural example has yet been found in India, and that all the arched or circular forms found in the caves are without one single exception copies of wooden forms, and nowhere even simulate stone construction. With the Hindus and Jains the case is different: they use stone arches and stone domes which are not copied from wooden forms at all; but these are invariably horizontal arches, never formed or intended to be formed with radiating voussoirs. It has already been explained, in speaking of Pelasgic art,[251] how prevalent these forms were in ancient Greece and Asia Minor, and how long they continued to be employed even after the principles of the true arch were perfectly understood. In India, however, the adherence to this form of construction is even more remarkable. As the Hindus quaintly express it, “an arch never sleeps;” and it is true that a radiating arch does contain in itself a _vis viva_ which is always tending to thrust its haunches outwards, and goes far to ensure the ultimate destruction of every building where it is employed: while the horizontal forms employed by the Hindus are in stable equilibrium, and, unless disturbed by violence, might remain so for ever. There can be no doubt that the Hindus carried their horror of an arch to an excess which frequently led them to worse faults on the other side. In city walls for instance, where there is a superabundant abutment on either hand to counteract any thrust, the horizontal principle is entirely misplaced. If we take, for instance, one of the city gates at Bijanagur (Woodcut No. 108), we cannot help perceiving that with much smaller stones and less trouble a far more stable construction could have been obtained, so long as the wall on either hand remained entire. What the Hindu feared was that if the wall were shattered, as we now find it, the arch would have fallen, though the horizontal layers still remain in their places. [Illustration: 108. View of City Gateway, Bijanagur. (From a Photograph.)] [Illustration: 109. Gateway, Jinjûwarra. (From Kinloch Forbes’ ‘Ras Mala.’)] Instead of a continuous bracket like that shown in the last example, a more usual form, in modern times at least, is that of several detached brackets placed a little distance apart the one from the other. When used in moderation this is the more pleasing form of the two, and in southern India it is generally used with great success. In the north they are liable to exaggerate it, as in the gateway from Jinjûwarra in Gujerat (Woodcut No. 109, p. 211), when it becomes unpleasing, though singularly characteristic of the style.