No. 17, which is very similar in plan, is generally known as the Zodiac cave, from the figure of a Buddhist chakra or wheel painted at one end of its verandah, which was mistaken by early visitors for a celestial emblem. The general effect of its architecture internally may be gathered from the above woodcut (No. 85) from a photograph, or from the next woodcut (No. 86) representing one of its pillars to a larger scale, from which the curiously wooden construction of the roof will be better observed than from the photograph. It is, in fact, the usual mode of forming flat or terraced roofs at the present day throughout India, and which consequently does not seem to have [Illustration: 86. Pillar in Vihara No. 17, at Ajunta. (From a Sketch by the Author.)] varied from the 5th century at all events. As may be gathered from these illustrations the pillars in these caves are almost indefinitely varied, generally in pairs, but no pillars in any one cave are at all like those in any other. In each cave, however, there is a general harmony of design and of form, which prevents their variety from being unpleasing. The effect on the contrary is singularly harmonious and satisfactory. The great interest of these two caves lies, however, in their frescoes, which represent Buddhist legends on a scale and with a distinctness found nowhere else in India. The sculptures of Amravati--some of which may be contemporary, or only slightly earlier--are what most nearly approach them; but, as in most cases, painting admits of greater freedom and greater variety of incident than sculpture ever does, and certainly in this instance vindicates its claim to greater phonetic power. Many of the frets and architectural details painted on the roofs and pillars of these and in viharas are also of great elegance and appropriateness, and, when combined with the architecture, make up a whole unrivalled in India for its ethnographic as well as for its architectural beauty. Fortunately the age of these two caves is not doubtful; there is a long inscription on each, much mutilated it must be confessed, but of which enough can be made out to show that they were excavated by kings of the Vindhyasacti race, one of whom, Pravarasena, whose name appears in the inscription on No. 16, married a daughter of Maharaja Deva, alias Chandra-gupta.[191] We have inscriptions of the last king dated 82 and 93 of the Gupta era, or in A.D. 400 and 411, and his son-in-law may probably have reigned a few years later. We may consequently safely place these two caves in the first half of the 5th century. They are thus slightly more modern than the Yadnya Sri cave, No. 15, at Nassick, which is exactly the result we would expect to arrive at from their architecture and the form of their sanctuaries.